blob: 96c5cc65026bcb938d58e478648c786a737b91f3 [file] [log] [blame]
<?xml version="1.0"?>
<?xml-stylesheet type="text/xsl" href="styleguide.xsl"?>
<GUIDE title="Google Common Lisp Style Guide">
<p align="right">
Revision 1.28
</p>
<address>
Robert Brown
</address>
<address>
<a HREF="mailto:tunes@google.com">François-René Rideau</a>
</address>
<address>
In memoriam Dan Weinreb
</address>
<p align="center">
<cite>Patterns mean "I have run out of language."</cite> — Rich Hickey
</p>
<OVERVIEW>
<CATEGORY title="Important Note">
<STYLEPOINT title="Note: Displaying Hidden Details in this Guide">
<SUMMARY>
This style guide contains many details
that are initially hidden from view.
They are marked by the triangle icon, which you see here on your left.
Click it now. You should see "Hooray" appear below.
</SUMMARY>
<BODY>
<p>
Hooray! Now you know you can expand points to get more details.
Alternatively, there's an "expand all" at the top of this document.
</p>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Background">
<p>
Common Lisp is a powerful multiparadigm programming language.
With great power comes great responsibility.
</p>
<p>
This guide recommends formatting and stylistic choices
designed to make your code easier for other people to understand.
For those internal applications and free software libraries that
we develop at Google,
you should keep within these guidelines when making changes.
Note however that each project has its own rules and customs
that complement or override these general guidelines;
the speed-oriented QPX low fare search engine notably
has a very different style and feel from the QRes reservation system.
</p>
<p>
If you're writing Common Lisp code outside Google,
we invite you to consider these guidelines.
You may find some of them useful
where they don't conflict with other priorities you have.
We welcome remarks and constructive feedback
on how to improve our guide, and
on what alternate styles work for you and why.
</p>
<p>
This guide is not a Common Lisp tutorial.
For basic information about the language, please consult
<a HREF="http://www.gigamonkeys.com/book/">Practical Common Lisp</a>.
For a language reference, please consult the
<a HREF="http://www.lispworks.com/documentation/HyperSpec/Front/index.htm">Common Lisp HyperSpec</a>.
For more detailed style guidance, take (with a pinch of salt)
a look at Peter Norvig and Kent Pitman's
<a HREF="http://norvig.com/luv-slides.ps">style guide</a>.
</p>
</CATEGORY>
</OVERVIEW>
<CATEGORY title="Meta-Guide">
<STYLEPOINT title="Must, Should, May, or Not">
<SUMMARY>
Each guideline's level of importance is indicated
by use of the following keywords and phrases, adapted from
<a href="https://www.ietf.org/rfc/rfc2119.txt">RFC 2119</a>.
</SUMMARY>
<BODY>
<table>
<tr>
<th valign="top">MUST</th>
<td>
<p>
This word, or the terms "REQUIRED" or "SHALL",
means that the guideline is an absolute requirement.
You must ask permission to violate a MUST.
</p>
</td>
</tr>
<tr>
<th valign="top">MUST NOT</th>
<td>
<p>
This phrase, or the phrase "SHALL NOT",
means that the guideline is an absolute prohibition.
You must ask permission to violate a MUST NOT.
</p>
</td>
</tr>
<tr>
<th valign="top">SHOULD</th>
<td>
<p>
This word, or the adjective "RECOMMENDED", means that
there may exist valid reasons in particular circumstances
to ignore the demands of the guideline, but
the full implications must be understood and carefully weighed
before choosing a different course.
You must ask forgiveness for violating a SHOULD.
</p>
</td>
</tr>
<tr>
<th valign="top">SHOULD NOT</th>
<td>
<p>
This phrase, or the phrase "NOT RECOMMENDED", means that
there may exist valid reasons in particular circumstances
to ignore the prohibitions of this guideline, but
the full implications should be understood and carefully weighed
before choosing a different course.
You must ask forgiveness for violating a SHOULD NOT.
</p>
</td>
</tr>
<tr>
<th valign="top">MAY</th>
<td>
<p>
This word, or the adjective "OPTIONAL",
means that an item is truly optional.
</p>
</td>
</tr>
</table>
<p>
Unlike RFCs, we don't capitalize every instance of one of the above
keywords when it is used.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Permission and Forgiveness">
<SUMMARY>
There are cases where transgression of some of these rules
is useful or even necessary.
In some cases, you must seek permission or obtain forgiveness
from the proper people.
</SUMMARY>
<BODY>
<p>
Permission comes from the owners of your project.
</p>
<p>
Forgiveness is requested in a comment
near the point of guideline violation,
and is granted by your code reviewer.
The original comment should be signed by you, and
the reviewer should add a signed approval to the comment at review time.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Conventions">
<SUMMARY>
You MUST follow conventions. They are not optional.
</SUMMARY>
<BODY>
<p>
Some of these guidelines are motivated by universal principles of good programming.
Some guidelines are motivated by technical peculiarities of Common Lisp.
Some guidelines were once motivated by a technical reason,
but the guideline remained after the reason subsided.
Some guidelines, such those about as comments and indentation,
are based purely on convention, rather than on clear technical merit.
Whatever the case may be, you must still follow these guidelines,
as well as other conventional guidelines
that have not been formalized in this document.
</p>
<p>
You MUST follow conventions.
They are important for readability.
When conventions are followed by default,
violations of the convention are a signal
that something notable is happening and deserves attention.
When conventions are systematically violated,
violations of the convention are a distracting noise
that needs to be ignored.
</p>
<p>
Conventional guidelines <em>are</em> indoctrination.
Their purpose is to make you follow the mores of the community,
so you can more effectively cooperate with existing members.
It is still useful to distinguish the parts that are technically motivated
from the parts that are mere conventions,
so you know when best to defy conventions for good effect,
and when not to fall into the pitfalls that the conventions are there to help avoid.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Old Code">
<SUMMARY>
Fix old code as you go.
</SUMMARY>
<BODY>
<p>
A lot of our code was written before these guidelines existed.
You should fix violations as you encounter them
in the course of your normal coding.
</p>
<p>
You must not fix violations en masse
without warning other developers and coordinating with them,
so as not to make the merging of large branches
more difficult than it already is.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Future Topics">
<SUMMARY>
There are many topics for additional standardization
not covered by current version of this document,
but deferred to future versions.
</SUMMARY>
<BODY>
<ul>
<li>
File and directory structure
</li>
<li>
Packages and modularity
</li>
<li>
Threads and locking
</li>
<li>
How to add configurable components
</li>
<li>
CLOS style: initforms, slot and accessor names, etc.
</li>
<li>
Recommendations on max number of slots per class.
</li>
<li>
More concrete examples of good code:
<ul>
<li>
exceptions
</li>
<li>
transactions, with retry
</li>
<li>
XML
</li>
<li>
typing
</li>
<li>
encapsulation / abstraction
</li>
<li>
class and slot names
</li>
<li>
etc.
</li>
</ul>
</li>
<li>
When (not) to use conditional compilation:
<ul>
<li>
modifying the product
</li>
<li>
conditional debugging/console output/etc.
</li>
<li>
"temporarily" commenting-out blocks of code
</li>
<li>
etc.
</li>
</ul>
</li>
</ul>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="General Guidelines">
<STYLEPOINT title="Principles">
<SUMMARY>
There are some basic principles for team software development
that every developer must keep in mind.
Whenever the detailed guidelines are inadequate, confusing or contradictory,
refer back to these principles for guidance:
<ul>
<li>
Every developer's code must be easy for another developer
to read, understand, and modify
— even if the first developer isn't around to explain it.
(This is the "hit by a truck" principle.)
</li>
<li>
Everybody's code should look the same.
Ideally, there should be no way to look at lines of code
and recognize it as "Fred's code" by its style.
</li>
<li>
Be precise.
</li>
<li>
Be concise.
</li>
<li>
KISS — Keep It Simple, Stupid.
</li>
<li>
Use the smallest hammer for the job.
</li>
<li>
Use common sense.
</li>
<li>
Keep related code together.
Minimize the amount of jumping around
someone has to do to understand an area of code.
</li>
</ul>
</SUMMARY>
<BODY>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Priorities">
<SUMMARY>
<p>
When making decisions about how to write a given piece of
code, aim for the following -ilities in this priority order:
</p>
<ul>
<li>
Usability by the customer
</li>
<li>
Debuggability/Testability
</li>
<li>
Readability/Comprehensibility
</li>
<li>
Extensibility/Modifiability
</li>
<li>
Efficiency (of the Lisp code at runtime)
</li>
</ul>
</SUMMARY>
<BODY>
<p>
Most of these are obvious.
</p>
<p>
Usability by the customer means that the system has to do what the
customer requires; it has to handle the customer's transaction
volumes, uptime requirements; etc.
</p>
<p>
For the Lisp efficiency point,
given two options of equivalent complexity,
pick the one that performs better.
(This is often the same as the one that conses less,
i.e. allocates less storage from the heap.)
</p>
<p>
Given two options where one is more complex than the other,
pick the simpler option and revisit the decision only if
profiling shows it to be a performance bottleneck.
</p>
<p>
However, avoid premature optimization.
Don't add complexity to speed up something that runs rarely,
since in the long run, it matters less whether such code is fast.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Architecture">
<SUMMARY>
To build code that is robust and maintainable,
it matters a lot how the code is divided into components,
how these components communicate,
how changes propagate as they evolve,
and more importantly
how the programmers who develop these components communicate
as these components evolve.
</SUMMARY>
<BODY>
<p>
If your work affects other groups, might be reusable across groups,
adds new components, has an impact on other groups
(including QA or Ops), or otherwise isn't purely local,
you must write it up using at least a couple of paragraphs,
and get a design approval from the other parties involved
before starting to write code — or be ready to scratch what you have
when they object.
</p>
<p>
If you don't know or don't care about these issues,
ask someone who does.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Using Libraries">
<SUMMARY>
Often, the smallest hammer is to use an existing library.
Or one that doesn't exist yet.
In such cases, you are encouraged to use or develop such a library,
but you must take appropriate precautions.
</SUMMARY>
<BODY>
<ul>
<li>
You MUST NOT start a new library
unless you established that none is already available
that can be fixed or completed into becoming what you need.
That's a rule against the NIH syndrome ("Not Invented Here"),
which is particularly strong amongst Lisp hackers.
</li>
<li>
Whichever library, old or new, you pick, you MUST get permission
to incorporate third-party code into the code base.
You must discuss the use of such library
in the appropriate mailing-list,
and have your code reviewed by people knowledgeable in the domain
and/or the Lisp library ecosystem (if any).
Please be ready to argue why this particular solution makes sense
as compared to other available libraries.
</li>
<li>
Some libraries are distributed under licenses not compatible
with the software you're writing, and
must not be considered available for use.
Be aware of these issues, or consult with people who are.
</li>
</ul>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Open-Sourcing Code">
<SUMMARY>
<p>
If you write a general-purpose library,
or modify an existing open-source library,
you are encouraged to publish the result
separate from your main project and then
have your project import it like any other open-source library.
</p>
</SUMMARY>
<BODY>
<p>
Use your judgment to distinguish
general-purpose versus business-specific code,
and open-source the general-purpose parts,
while keeping the business-specific parts a trade secret.
</p>
<p>
Open-Sourcing code has many advantages,
including being able to leverage third parties for development,
letting the development of features be user-directed,
and keeping you honest with respect to code quality.
Whatever code you write, you will have to maintain anyway,
and make sure its quality is high enough to sustain use in production.
There should therefore be no additional burden to Open-Sourcing,
even of code that (at least initially)
is not directly usable by third parties.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Development Process">
<SUMMARY>
Development process is outside the scope of this document.
However, developers should remember at least these bits:
get reviewed, write tests, eliminate warnings, run tests, avoid mass-changes.
</SUMMARY>
<BODY>
<p>
</p>
<ul>
<li>
All code changes must be reviewed.
You should expect that your code will be reviewed by other hackers,
and that you will be assigned other hackers' code to review.
Part of the review criteria will be that code obeys
the coding standards in this document.
</li>
<li>
You must write and check-in tests for new code that you write and old bugs you fix.
There must be a unit test for every API function,
and any previously failing case.
Your work is not truly done until this activity is complete.
Estimating tasks must include the time it takes to produce such tests.
</li>
<li>
Your code must compile
without any compilation error or warning messages whatsoever.
If the compiler issues warnings that should be ignored,
muffle those warnings using the
<code>UIOP:WITH-MUFFLED-COMPILER-CONDITIONS</code> and
<code>UIOP:*UNINTERESTING-COMPILER-CONDITIONS*</code>
framework (part of <code>UIOP</code>, part of <code>ASDF 3</code>),
either around the entire project, or around individual files
(using <code>ASDF</code>'s <code>:around-compile</code> hooks).
</li>
<li>
All code should be checked in an appropriate source control system,
in a way that allows for complete reproducibility of
build, test and execution of
the code that is, has been or may be deployed.
</li>
<li>
You must run the "precheckin" tests, and each component must pass
its unit tests successfully before you commit any code.
</li>
<li>
You should incorporate code coverage into your testing process.
Tests are not sufficient
if they do not cover all new and updated code;
code that for whatever reason cannot be included in coverage results
should be clearly marked as such including the reason.
</li>
<li>
Many people develop on branches.
You must get permission to undertake mass-changes
(e.g. mass reindentations)
so that we can coordinate in advance,
and give branch residents time to get back on the mainline
</li>
</ul>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Formatting">
<STYLEPOINT title="Spelling and Abbreviations">
<SUMMARY>
<p>
You must use correct spelling in your comments,
and most importantly in your identifiers.
</p>
<p>
When several correct spellings exist (including American vs English),
and there isn't a consensus amongst developers as which to use,
you should choose the shorter spelling.
</p>
<p>
You must use only common and domain-specific abbreviations, and
must be consistent with these abbreviations. You may abbreviate
lexical variables of limited scope in order to avoid overly-long
symbol names.
</p>
</SUMMARY>
<BODY>
<p>
If you're not sure, consult a dictionary,
Google for alternative spellings,
or ask a local expert.
</p>
<p>
Here are examples of choosing the correct spelling:
</p>
<ul>
<li>
Use "complimentary" in the sense of a meal or beverage
that is not paid for by the recipient, not "complementary".
</li>
<li>
Use "existent" and "nonexistent", not "existant".
Use "existence", not "existance".
</li>
<li>
Use "hierarchy" not "heirarchy".
</li>
<li>
Use "precede" not "preceed".
</li>
<li>
Use "weird", not "wierd".
</li>
</ul>
<p>
Here are examples of choosing the shorter spelling:
</p>
<ul>
<li>
Use "canceled", not "cancelled"
</li>
<li>
Use "queuing", not "queueing".
</li>
<li>
Use "signaled", not "signalled";
</li>
<li>
Use "traveled", not "travelled".
</li>
<li>
Use "aluminum", not "aluminium"
</li>
<li>
Use "oriented", not "orientated"
</li>
<li>
Use "color", not "colour"
</li>
<li>
Use "behavior", not "behaviour"
</li>
</ul>
<p>
Make appropriate exceptions for industry standard nomenclature/jargon,
including plain misspellings.
For instance:
</p>
<ul>
<li>
Use "referer", not "referrer", in the context of the HTTP protocol.
</li>
</ul>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Line length">
<SUMMARY>
You should format source code so that no line is longer than 100 characters.
</SUMMARY>
<BODY>
<p>
Some line length restriction is better than none at all.
While old text terminals used to make 80 columns the standard,
these days, allowing 100 columns seems better,
since good style encourages the use of
descriptive variables and function names.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Indentation">
<SUMMARY>
<p>
Indent your code the way a properly configured GNU Emacs does.
</p>
<p>
Maintain a consistent indentation style throughout a project.
</p>
<p>
Indent carefully to make the code easier to understand.
</p>
</SUMMARY>
<BODY>
<p>
Common Lisp indentation in Emacs is provided by the cl-indent library.
The latest version of cl-indent is packaged with
<a HREF="https://www.common-lisp.net/project/slime/">SLIME</a>
(under contrib/slime-cl-indent.el). After installing SLIME, set up Emacs
to load SLIME automatically using
<a HREF="https://www.common-lisp.net/project/slime/doc/html/Loading-Contribs.html">these instructions</a>, adding slime-indentation to the list of
contrib libraries to be loaded in the call to slime-setup.
</p>
<p>
Ideally, use the default indentation settings provided by
slime-indentation. If necessary, customize indentation parameters to
maintain a consistent indentation style throughout an existing project.
Parameters can be customized using the :variables setting in
define-common-lisp-style. Indentation of specific forms can be
customized using the :indentation setting of define-common-lisp-style.
This is particularly useful when creating forms that behave like macros
or special operators that are indented differently than standard
function calls (e.g. defun, labels, or let). Add a
<a HREF="https://www.gnu.org/software/emacs/manual/html_node/emacs/Hooks.html">hook</a> to 'lisp-mode-hook that calls common-lisp-set-style to set
the appropriate style automatically.
</p>
<p>
Use indentation to make complex function applications easier to read.
When an application does not fit on one line
or the function takes many arguments,
consider inserting newlines between the arguments
so that each one is on a separate line.
However, do not insert newlines in a way that makes it hard to tell
how many arguments the function takes
or where an argument form starts and ends.
</p>
<BAD_CODE_SNIPPET>
;; Bad
(do-something first-argument second-argument (lambda (x)
(frob x)) fourth-argument last-argument)
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
;; Better
(do-something first-argument
second-argument
#'(lambda (x) (frob x))
fourth-argument
last-argument)
</CODE_SNIPPET>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="File Header">
<SUMMARY>
<p>
You should include a description at the top of each source file.
</p>
<p>
You should include neither authorship nor copyright information in a source file.
</p>
</SUMMARY>
<BODY>
<p>
Every source file should begin with a brief description
of the contents of that file.
</p>
<p>
After that description, every file should start the code itself with an
<code>(in-package #:<em>package-name</em>)</code> form.
</p>
<p>
After that <code>in-package</code> form,
every file should follow with any file-specific
<code>(declaim (optimize ...))</code> declaration
that is not covered by an <code>ASDF</code> <code>:around-compile</code> hook.
</p>
<CODE_SNIPPET>
;;;; Variable length encoding for integers and floating point numbers.
(in-package #:varint)
(declaim #.*optimize-default*)
</CODE_SNIPPET>
<p>
You should not include authorship information at the top of a file:
better information is available from version control,
and such a mention will only cause confusion and grief.
Indeed, whoever was the main author at the time such a mention was included
might not be who eventually made the most significant contributions to the file,
and even less who is responsible for the file at the moment.
</p>
<p>
You should not include copyright information in individual source code files.
An exception is made for files meant to be disseminated as standalone.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Vertical white space">
<SUMMARY>
Vertical white space: one blank line between top-level forms.
</SUMMARY>
<BODY>
<p>
You should include one blank line between top-level forms,
such as function definitions.
Exceptionally, blank lines can be omitted
between simple, closely related defining forms of the same kind,
such as a group of related type declarations or constant definitions.
</p>
<CODE_SNIPPET>
(defconstant +mix32+ #x12b9b0a1 "pi, an arbitrary number")
(defconstant +mix64+ #x2b992ddfa23249d6 "more digits of pi")
(defconstant +golden-ratio32+ #x9e3779b9 "the golden ratio")
(defconstant +golden-ratio64+ #xe08c1d668b756f82 "more digits of the golden ratio")
(defmacro incf32 (x y)
"Like INCF, but for integers modulo 2**32"
`(setf ,x (logand (+ ,x ,y) #xffffffff)))
(defmacro incf64 (x y)
"Like INCF, but for integers modulo 2**64"
`(setf ,x (logand (+ ,x ,y) #xffffffffffffffff)))
</CODE_SNIPPET>
<p>
Blank lines can be used to separate parts of a complicated function.
Generally, however, you should break a large function into smaller ones
instead of trying to make it more readable by adding vertical space.
If you can't, you should document with a <code>;;</code> comment
what each of the separated parts of the function does.
</p>
<p>
You should strive to keep top-level forms,
including comments but excluding the documentation string, of
appropriate length; preferrably short. Forms extending beyond a
single page should be rare and their use should be justfied.
This applies to each of the forms in an <code>eval-when</code>,
rather than to the <code>eval-when</code> itself.
Additionally, <code>defpackage</code> forms may be longer,
since they may include long lists of symbols.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Horizontal white space">
<SUMMARY>
Horizontal white space: none around parentheses. No tabs.
</SUMMARY>
<BODY>
<p>
You must not include extra horizontal whitespace
before or after parentheses or around symbols.
</p>
<p>
You must not place right parentheses by themselves on a line.
A set of consecutive trailing parentheses must appear on the same line.
</p>
<BAD_CODE_SNIPPET>
;; Very Bad
( defun factorial ( limit )
( let (( product 1 ))
( loop for i from 1 upto limit
do (setf product ( * product i ) ) )
product
)
)
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
;; Better
(defun factorial (limit)
(let ((product 1))
(loop for i from 1 upto limit
do (setf product (* product i)))
product))
</CODE_SNIPPET>
<p>
You should use only one space between forms.
</p>
<p>
You should not use spaces to vertically align forms
in the middle of consecutive lines.
An exception is made when the code possesses
an important yet otherwise not visible symmetry
that you want to emphasize.
</p>
<BAD_CODE_SNIPPET>
;; Bad
(let* ((low 1)
(high 2)
(sum (+ (* low low) (* high high))))
...)
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
;; Better
(let* ((low 1)
(high 2)
(sum (+ (* low low) (* high high))))
...))
</CODE_SNIPPET>
<p>
You must align nested forms if they occur across more than one line.
</p>
<BAD_CODE_SNIPPET>
;; Bad
(defun munge (a b c)
(* (+ a b)
c))
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
;; Better
(defun munge (a b c)
(* (+ a b)
c))
</CODE_SNIPPET>
<p>
The convention is that the body of a binding form
is indented two spaces after the form.
Any binding data before the body is usually indented four spaces.
Arguments to a function call are aligned with the first argument;
if the first argument is on its own line,
it is aligned with the function name.
</p>
<CODE_SNIPPET>
(multiple-value-bind (a b c d)
(function-returning-four-values x y)
(declare (ignore c))
(something-using a)
(also-using b d))
</CODE_SNIPPET>
<p>
An exception to the rule against lonely parentheses
is made for an <code>eval-when</code> form around several definitions;
in this case, include a comment <code>; eval-when</code>
after the closing parenthesis.
</p>
<p>
You must set your editor to
avoid inserting tab characters in the files you edit.
Tabs cause confusion when editors disagree
on how many spaces they represent.
In Emacs, do <code>(setq-default indent-tabs-mode nil)</code>.
</p>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Documentation">
<STYLEPOINT title="Document everything">
<SUMMARY>
You should use document strings on all visible functions
to explain how to use your code.
</SUMMARY>
<BODY>
<p>
Unless some bit of code is painfully self-explanatory,
document it with a documentation string (also known as docstring).
</p>
<p>
Documentation strings are destined to be read
by the programmers who use your code.
They can be extracted from functions, types, classes, variables and macros,
and displayed by programming tools, such as IDEs,
or by REPL queries such as <code>(describe 'foo)</code>;
web-based documentation or other reference works
can be created based on them.
Documentation strings are thus the perfect locus to document your API.
They should describe how to use the code
(including what pitfalls to avoid),
as opposed to how the code works (and where more work is needed),
which is what you'll put in comments.
</p>
<p>
Supply a documentation string when defining
top-level functions, types, classes, variables and macros.
Generally, add a documentation string wherever the language allows.
</p>
<p>
For functions, the docstring should describe the function's contract:
what the function does,
what the arguments mean,
what values are returned,
what conditions the function can signal.
It should be expressed at the appropriate level of abstraction,
explaining the intended meaning rather than, say, just the syntax.
In documentation strings, capitalize the names of Lisp symbols,
such as function arguments.
For example, "The value of LENGTH should be an integer."
</p>
<CODE_SNIPPET>
(defun small-prime-number-p (n)
"Return T if N, an integer, is a prime number. Otherwise, return NIL."
(cond ((or (&lt; n 2))
nil)
((= n 2)
t)
((divisorp 2 n)
nil)
(t
(loop for i from 3 upto (sqrt n) by 2
never (divisorp i n)))))
</CODE_SNIPPET>
<CODE_SNIPPET>
(defgeneric table-clear (table)
(:documentation
"Like clrhash, empties the TABLE of all
associations, and returns the table itself."))
</CODE_SNIPPET>
<p>
A long docstring may usefully
begin with a short, single-sentence summary,
followed by the larger body of the docstring.
</p>
<p>
When the name of a type is used,
the symbol may be quoted by surrounding it with
a back quote at the beginning and a single quote at the end.
Emacs will highlight the type, and the highlighting serves
as a cue to the reader that <kbd>M-.</kbd>
will lead to the symbol's definition.
</p>
<CODE_SNIPPET>
(defun bag-tag-expected-itinerary (bag-tag)
"Return a list of `legacy-pnr-pax-segment' objects representing
the expected itinerary of the `bag-tag' object, BAG-TAG."
...)
</CODE_SNIPPET>
<p>
Every method of a generic function should be independently documented
when the specialization affects what the method does,
beyond what is described in its generic function's docstring.
</p>
<p>
When you fix a bug,
consider whether what the fixed code does is obviously correct or not;
if not, you must add a comment explaining
the reason for the code in terms of fixing the bug.
Adding the bug number, if any, is also recommended.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Comment semicolons">
<SUMMARY>
You must use the appropriate number of semicolons to introduce comments.
</SUMMARY>
<BODY>
<p>
Comments are explanations to the future maintainers of the code.
Even if you're the only person who will ever see and touch the code,
even if you're either immortal and never going to quit,
or unconcerned with what happens after you leave
(and have your code self-destruct in such an eventuality),
you may find it useful to comment your code.
Indeed, by the time you revisit your code,
weeks, months or years later,
you will find yourself a different person from the one who wrote it,
and you will be grateful to that previous self
for making the code readable.
</p>
<p>
You must comment anything complicated
so that the next developer can understand what's going on.
(Again, the "hit by a truck" principle.)
</p>
<p>
Also use comments as a way to guide those who read the code,
so they know what to find where.
</p>
<ul>
<li>
File headers and important comments
that apply to large sections of code in a source file
should begin with four semicolons.
</li>
<li>
You should use three semicolons
to begin comments that apply to just
one top-level form or small group of top-level forms.
</li>
<li>
Inside a top-level form, you should use two semicolons
to begin a comment if it appears between lines.
</li>
<li>
You should use one semicolon if it is a parenthetical remark
and occurs at the end of a line.
You should use spaces to separate the comment
from the code it refers to so the comment stands out.
You should try to vertically align
consecutive related end-of-line comments.
</li>
</ul>
<CODE_SNIPPET>
;;;; project-euler.lisp
;;;; File-level comments or comments for large sections of code.
;;; Problems are described in more detail here: https://projecteuler.net/
;;; Divisibility
;;; Comments that describe a group of definitions.
(defun divisorp (d n)
(zerop (mod n d)))
(defun proper-divisors (n)
...)
(defun divisors (n)
(cons n (proper-divisors n)))
;;; Prime numbers
(defun small-prime-number-p (n)
(cond ((or (&lt; n 2))
nil)
((= n 2) ; parenthetical remark here
t) ; continuation of the remark
((divisorp 2 n)
nil) ; different remark
;; Comment that applies to a section of code.
(t
(loop for i from 3 upto (sqrt n) by 2
never (divisorp i n)))))
</CODE_SNIPPET>
<p>
You should include a space between the semicolon and the text of the comment.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Grammar and punctuation">
<SUMMARY>
You should punctuate documentation correctly.
</SUMMARY>
<BODY>
<p>
When a comment is a full sentence,
you should capitalize the initial letter of the first word
and end the comment with a period.
In general, you should use correct punctuation.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Attention Required">
<SUMMARY>
You must follow the convention of using TODO comments
for code requiring special attention.
For code using unobvious forms, you must include a comment.
</SUMMARY>
<BODY>
<p>
For comments requiring special attention, such as
incomplete code, todo items, questions, breakage, and danger,
include a TODO comment indicating the type of problem,
its nature, and any notes on how it may be addressed.
</p>
<p>
The comments begin with <code>TODO</code> in all capital letters,
followed by the
name, e-mail address, or other identifier
of the person
with the best context about the problem referenced by the <code>TODO</code>.
The main purpose is to have a consistent <code>TODO</code> that
can be searched to find out how to get more details upon
request. A <code>TODO</code> is not a commitment that the
person referenced will fix the problem. Thus when you create
a <code>TODO</code>,
it is almost always your
name
that is given.
</p>
<p>
When signing comments,
you should use your username (for code within the company)
or full email address (for code visible outside the company),
not just initials.
</p>
<CODE_SNIPPET>
;;--- TODO(george@gmail.com): Refactor to provide a better API.
</CODE_SNIPPET>
<p>
Be specific when indicating times or software releases
in a TODO comment and use
<a HREF="https://www.w3.org/TR/NOTE-datetime">YYYY-MM-DD</a>
format for dates to make automated processing of such dates easier,
e.g., 2038-01-20 for the end of the 32-bit signed <code>time_t</code>.
</p>
<CODE_SNIPPET>
;;--- TODO(brown): Remove this code after release 1.7 or before 2012-11-30.
</CODE_SNIPPET>
<p>
For code that uses unobvious forms to accomplish a task, you must include a comment
stating the purpose of the form and the task it accomplishes.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Domain-Specific Languages">
<SUMMARY>
You should document DSLs and
any terse program in a DSL.
</SUMMARY>
<BODY>
<p>
You should design your Domain Specific Language
to be easy to read and understand by people familiar with the domain.
</p>
<p>
You must properly document all your Domain Specific Language.
</p>
<p>
Sometimes, your DSL is designed for terseness.
In that case, it is important to document what each program does,
if it's not painfully obvious from the context.
</p>
<p>
Notably, when you use regular expressions
(e.g. with the <code>CL-PPCRE</code> package),
you MUST ALWAYS put in a comment
(usually a two-semicolon comment on the previous line)
explaining, at least basically, what the regular expression does,
or what the purpose of using it is.
The comment need not spell out every bit of the syntax, but
it should be possible for someone to follow the logic of the code
without actually parsing the regular expression.
</p>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Naming">
<STYLEPOINT title="Symbol guidelines">
<SUMMARY>
You should use lower case.
You should follow the rules for <a href="#Spelling_and_Abbreviations">Spelling and Abbreviations</a>
You should follow punctuation conventions.
</SUMMARY>
<BODY>
<p>
Use lower case for all symbols.
Consistently using lower case makes searching for symbol names easier
and is more readable.
</p>
<p>
Note that Common Lisp is case-converting,
and that the <code>symbol-name</code> of your symbols
will be upper case.
Because of this case-converting,
attempts to distinguish symbols by case are defeated,
and only result in confusion.
While it is possible to escape characters in symbols
to force lower case,
you should not use this capability
unless this is somehow necessary
to interoperate with third-party software.
</p>
<p>
Place hyphens between all the words in a symbol.
If you can't easily say an identifier out loud,
it is probably badly named.
</p>
<p>
You must not use <code>"/"</code> or <code>"."</code>
instead of <code>"-"</code>
unless you have a well-documented overarching reason to,
and permission from other hackers who review your proposal.
</p>
<p>
See the section on <a href="#Spelling_and_Abbreviations">Spelling and Abbreviations</a>
for guidelines on using abbreviations.
</p>
<BAD_CODE_SNIPPET>
;; Bad
(defvar *default-username* "Ann")
(defvar *max-widget-cnt* 200)
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
;; Better
(defvar *default-user-name* "Ann")
(defvar *maximum-widget-count* 200)
</CODE_SNIPPET>
<p>
There are conventions in Common Lisp
for the use of punctuation in symbols.
You should not use punctuation in symbols outside these conventions.
</p>
<p>
Unless the scope of a variable is very small,
do not use overly short names like
<code>i</code> and <code>zq</code>.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Denote intent, not content">
<SUMMARY>
Name your variables according to their intent,
not their content.
</SUMMARY>
<BODY>
<p>
You should name a variable according
to the high-level concept that it represents,
not according to the low-level implementation details
of how the concept is represented.
</p>
<p>
Thus, you should avoid embedding
data structure or aggregate type names,
such as <code>list</code>, <code>array</code>, or
<code>hash-table</code> inside variable names,
unless you're writing a generic algorithm that applies to
arbitrary lists, arrays, hash-tables, etc.
In that case it's perfectly OK to name a variable
<code>list</code> or <code>array</code>.
</p>
<p>
Indeed, you should be introducing new abstract data types
with <code>DEFCLASS</code> or <code>DEFTYPE</code>,
whenever a new kind of intent appears for objects in your protocols.
Functions that manipulate such objects generically may then
use variables the name of which reflect that abstract type.
</p>
<p>
For example, if a variable's value is always a row
(or is either a row or <code>NIL</code>),
it's good to call it <code>row</code> or <code>first-row</code>
or something like that.
It is alright is <code>row</code> has been
<code>DEFTYPE</code>'d to <code>STRING</code>
precisely because you have abstracted the detail away,
and the remaining salient point is that it is a row.
You should not name the variable <code>STRING</code> in this context,
except possibly in low-level functions that specifically manipulate
the innards of rows to provide the suitable abstraction.
</p>
<p>
Be consistent.
If a variable is named <code>row</code> in one function,
and its value is being passed to a second function,
then call it <code>row</code> rather than, say, <code>value</code>
(this was a real case).
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Global variables and constants">
<SUMMARY>
Name globals according to convention.
</SUMMARY>
<BODY>
<p>
The names of global constants should start and end
with plus characters.
</p>
<p>
Global variable names should start and end with asterisks
(also known in this context as earmuffs).
</p>
<p>
In some projects, parameters that are not meant
to be usually modified or bound under normal circumstances
(but may be during experimentation or exceptional situations)
should start (but do not end) with a dollar sign.
If such a convention exists within your project,
you should follow it consistently.
Otherwise, you should avoid naming variables like this.
</p>
<p>
Common Lisp does not have global lexical variables,
so a naming convention is used to ensure that globals,
which are dynamically bound,
never have names that overlap with local variables.
It is possible to fake global lexical variables
with a differently named global variable
and a <code>DEFINE-SYMBOL-MACRO</code>.
You should not use this trick,
unless you first publish a library that abstracts it away.
</p>
<CODE_SNIPPET>
(defconstant +hash-results+ #xbd49d10d10cbee50)
(defvar *maximum-search-depth* 100)
</CODE_SNIPPET>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Predicate names">
<SUMMARY>
Names of predicate functions and variables end with a <code>"P"</code>.
</SUMMARY>
<BODY>
<p>
You should name boolean-valued functions and variables with a
trailing <code>"P"</code> or <code>"-P"</code>,
to indicate they are predicates.
Generally, you should use
<code>"P"</code> when the rest of the function name is one word
and <code>"-P"</code> when it is more than one word.
</p>
<p>
A rationale for this convention is given in
<a href="http://www.cs.cmu.edu/Groups/AI/html/cltl/clm/node69.html">the CLtL2 chapter on predicates</a>.
</p>
<p>
For uniformity, you should follow the convention above,
and not one of the alternatives below.
</p>
<p>
An alternative rule used in some existing packages
is to always use <code>"-P"</code>.
Another alternative rule used in some existing packages
is to always use <code>"?"</code>.
When you develop such a package,
you must be consistent with the rest of the package.
When you start a new package,
you should not use such an alternative rule
without a very good documented reason.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Omit library prefixes">
<SUMMARY>
You should not include a library or package name
as a prefix within the name of symbols.
</SUMMARY>
<BODY>
<p>
When naming a symbol (external or internal) in a package,
you should not include the package name
as a prefix within the name of the symbol.
Naming a symbol this way makes it awkward to use
from a client package accessing the symbol
by qualifying it with a package prefix,
where the package name then appears twice
(once as a package prefix,
another time as a prefix within the symbol name).
</p>
<BAD_CODE_SNIPPET>
;; Bad
(in-package #:varint)
(defun varint-length64 () ... )
(in-package #:client-code)
(defconst +padding+ (varint:varint-length64 +end-token+))
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
;; Better
(in-package #:varint)
(defun length64 () ... )
(in-package #:client-code)
(defconst +padding+ (varint:length64 +end-token+))
</CODE_SNIPPET>
<p>
An exception to the above rule would be to include a prefix
for the names of variables that would otherwise be expected to clash
with variables in packages that use the current one.
For instance, <code>ASDF</code> exports a variable <code>*ASDF-VERBOSE*</code>
that controls the verbosity of <code>ASDF</code> only,
rather than of the entire Lisp program.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Packages">
<SUMMARY>
Use packages appropriately.
</SUMMARY>
<BODY>
<p>
Lisp packages are used to demarcate namespaces.
Usually, each system has its own namespace.
A package has a set of external symbols,
which are intended to be used from outside the package,
in order to allow other modules to use this module's facilities.
</p>
<p>
The internal symbols of a package
should never be referred to from other packages.
That is, you should never have to use
the double-colon <code>::</code> construct.
(e.g. <code>QUAKE::HIDDEN-FUNCTION</code>).
If you need to use double-colons to write real production code,
something is wrong and needs to be fixed.
</p>
<p>
As an exception,
unit tests may use the internals of the package being tested.
So when you refactor, watch out for
internals used by the package's unit tests.
</p>
<p>
The <code>::</code> construct is also useful for very temporary hacks,
and at the REPL.
But if the symbol really is part of
the externally-visible definition of the package,
export it.
</p>
<p>
You may find that some internal symbols represent concepts
you usually want to abstract away and hide under the hood,
yet at times are necessary to expose for various extensions.
For the former reason, you do not want to export them,
yet for the latter reason, you have to export them.
The solution is to have two different packages,
one for your normal users to use, and
another for the implementation and its extenders to use.
</p>
<p>
Each package is one of two types:
</p>
<ul>
<li>
Intended to be included
in the <code>:use</code> specification of other packages.
If package <code>A</code> "uses" package <code>B</code>,
then the external symbols of package <code>B</code>
can be referenced from within package <code>A</code>
without a package prefix.
We mainly use this for low-level modules
that provide widely-used facilities.
</li>
<li>
Not intended to be "used".
To reference a facility provided by package <code>B</code>,
code in package <code>A</code> must use an explicit package prefix,
e.g. <code>B:DO-THIS</code>.
</li>
</ul>
<p>
If you add a new package, it should always be of the second type,
unless you have a special reason and get permission.
Usually a package is designed to be one or the other,
by virtue of the names of the functions.
For example, if you have an abstraction called <code>FIFO</code>,
and it were in a package of the first type
you'd have functions named things like
<code>FIFO-ADD-TO</code> and <code>FIFO-CLEAR-ALL</code>.
If you used a package of the second type,
you'd have names like <code>ADD-TO</code> and <code>CLEAR-ALL</code>,
because the callers would be saying
<code>FIFO:ADD-TO</code> and <code>FIFO:CLEAR-ALL</code>.
(<code>FIFO:FIFO-CLEAR-ALL</code> is redundant and ugly.)
</p>
<p>
Another good thing about packages is that
your symbol names won't "collide" with the names of other packages,
except the ones your packages "uses".
So you have to stay away from symbols
that are part of the Lisp implementation (since you always "use" that)
and that are part of any other packages you "use",
but otherwise you are free to make up your own names,
even short ones, and not worry about some else
having used the same name.
You're isolated from each other.
</p>
<p>
Your package must not shadow (and thus effectively redefine)
symbols that are part of the Common Lisp language.
There are certain exceptions,
but they should be very well-justified and extremely rare:
</p>
<ul>
<li>
If you are explicitly replacing a Common Lisp symbol
by a safer or more featureful version.
</li>
<li>
If you are defining a package not meant to be "used",
and have a good reason to export a symbol
that clashes with Common Lisp,
such as <code>log:error</code> and <code>log:warn</code>
and so on.
</li>
</ul>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Language usage guidelines">
<STYLEPOINT title="Mostly Functional Style">
<SUMMARY>
You should avoid side-effects when they are not necessary.
</SUMMARY>
<BODY>
<p>
Lisp is best used as a "mostly functional" language.
</p>
<p>
Avoid modifying local variables, try rebinding instead.
</p>
<p>
Avoid creating objects and the SETFing their slots.
It's better to set the slots during initialization.
</p>
<p>
Make classes as immutable as possible, that is, avoid giving slots
setter functions if at all possible.
</p>
<p>
Using a mostly functional style makes it much easier
to write concurrent code that is thread-safe.
It also makes it easier to test the code.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Recursion">
<SUMMARY>
You should favor iteration over recursion.
</SUMMARY>
<BODY>
<p>
Common Lisp systems are not required to implement
function calls from tail positions without leaking stack space
— which is known as proper tail calls (PTC),
tail call elimination (TCE),
or tail call optimization (TCO).
This means that indefinite recursion through tail calls
may quickly blow out the stack,
which hampers functional programming.
Still, most serious implementations (including SBCL and CCL)
do implement proper tail calls, but with restrictions:
</p>
<ul>
<li>
The <code>(DECLARE (OPTIMIZE ...))</code> settings
must favor <code>SPEED</code> enough and
not favor <code>DEBUG</code> too much,
for some compiler-dependent meanings of "enough" and "too much".
(For instance, in SBCL, you should avoid <code>(SPEED 0)</code>
and <code>(DEBUG 3)</code> to achieve proper tail calls.)
</li>
<li>
There should not be dynamic bindings around the call
(even though some Scheme compilers are able to properly treat
such dynamic bindings, called parameters in Scheme parlance).
</li>
</ul>
<p>
For compatibility with all compilers and optimization settings,
and to avoid stack overflow when debugging,
you should prefer iteration or the built in mapping functions
to relying on proper tail calls.
</p>
<p>
If you do rely on proper tail calls,
you must prominently document the fact,
and take appropriate measures to ensure an appropriate compiler is used
with appropriate optimization settings.
For fully portable code, you may have to use trampolines instead.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Special variables">
<SUMMARY>
Use special variables sparingly.
</SUMMARY>
<BODY>
<p>
Using Lisp "special" (dynamically bound) variables
as implicit arguments to functions should be used sparingly,
and only in cases where it won't surprise the person reading the code,
and where it offers significant benefits.
</p>
<p>
Indeed, each special variable constitutes state.
Developers have to mentally track the state of all relevant variables
when trying to understand what the code does and how it does it;
tests have to be written and run with all relevant combinations;
to isolate some activity, care has to be taken to locally bind
all relevant variables, including those of indirectly used modules.
They can hide precious information from being printed in a backtrace.
Not only is there overhead associated to each new variable,
but interactions between variables
can make the code exponentially more complex
as the number of such variables increases.
The benefits have to match the costs.
</p>
<p>
Note though that a Lisp special variable is not a global variable
in the sense of a global variable in, say, BASIC or C.
As special variables can be dynamically bound to a local value,
they are much more powerful than
global value cells where all users necessarily interfere with each other.
</p>
<p>
Good candidates for such special variables
are items for which "the current" can be naturally used as prefix,
such as "the current database connection" or
"the current business data source".
They are singletons as far as the rest of the code is concerned,
and often passing them as an explicit argument
does not add anything to the readability or maintainability
of the source code in question.
</p>
<p>
They can make it easier to write code that can be refactored.
If you have a request processing chain,
with a number of layers that all operate upon a "current" request,
passing the request object explicitly to every function
requires that every function in the chain have a request argument.
Factoring out code into new functions often requires
that these functions also have this argument,
which clutters the code with boilerplate.
</p>
<p>
You should treat special variables
as though they are per-thread variables.
By default, you should leave a special variable
with no top-level binding at all,
and each thread of control
that needs the variable should bind it explicitly.
This will mean that any incorrect use of the variable
will result in an "unbound variable" error, and
each thread will see its own value for the variable.
Variables with a default global value should usually be
locally bound at thread creation time.
You should use suitable infrastructure
to automate the appropriate declaration of such variables.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Assignment">
<SUMMARY>
Be consistent in assignment forms.
</SUMMARY>
<BODY>
<p>
There are several styles for dealing with assignment and side-effects;
whichever a given package is using,
keep using the same consistently when hacking said package.
Pick a style that makes sense when starting a new package.
</p>
<p>
Regarding multiple assignment in a same form, there are two schools:
the first style groups as many assignments as possible into a single
<code>SETF</code> or <code>PSETF</code> form
thus minimizing the number of forms with side-effects;
the second style splits assignments into as many individual
<code>SETF</code> (or <code>SETQ</code>, see below) forms as possible,
to maximize the chances of locating forms that modify a kind of place
by grepping for <code>(setf (foo ...</code>.
A grep pattern must actually contain as many place-modifying forms
as you may use in your programs, which may make this rationale either
convincing or moot depending on the rest of the style of your code.
You should follow the convention used in the package you are hacking.
We recommend the first convention for new packages.
</p>
<p>
Regarding <code>SETF</code> and <code>SETQ</code>,
there are two schools:
this first regards <code>SETQ</code>
as an archaic implementation detail,
and avoids it entirely in favor of <code>SETF</code>;
the second regards <code>SETF</code>
as an additional layer of complexity,
and avoids it in favor of <code>SETQ</code> whenever possible
(i.e. whenever the assigned place is a variable or symbol-macro).
You should follow the convention used in the package you are hacking.
We recommend the first convention for new packages.
</p>
<p>
In the spirit of a mostly pure functional style,
which makes testing and maintenance easier,
we invite you to consider how to do things with the fewest assignments required.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Assertions and Conditions">
<SUMMARY>
You must make proper usage of assertions and conditions.
</SUMMARY>
<BODY>
<ul>
<li>
<code>ASSERT</code> should be used ONLY to detect internal bugs.
Code should <code>ASSERT</code> invariants whose failure indicates
that the software is itself broken.
Incorrect input should be handled properly at runtime,
and must not cause an assertion violation.
The audience for an <code>ASSERT</code> failure is a developer.
Do not use the data-form and argument-form in <code>ASSERT</code>
to specify a condition to signal.
It's fine to use them to print out a message for debugging purposes
(and since it's only for debugging, there's no issue of
internationalization).
</li>
<li>
<code>CHECK-TYPE</code>,
<code>ETYPECASE</code> are also forms of assertion.
When one of these fails, that's a detected bug.
You should prefer to use <code>CHECK-TYPE</code>
over (DECLARE (TYPE ...))
for the inputs of functions.
</li>
<li>
Your code should use assertions and type checks liberally.
The sooner a bug is discovered, the better!
Only code in the critical path for performance
and internal helpers should eschew
explicit assertions and type checks.
</li>
<li>
Invalid input, such as files that are read
but do not conform to the expected format,
should not be treated as assertion violations.
Always check to make sure that input is valid,
and take appropriate action if it is not,
such as signalling a real error.
</li>
<li>
<code>ERROR</code> should be used
to detect problems with user data, requests, permissions, etc.,
or to report "unusual outcomes" to the caller.
</li>
<li>
<code>ERROR</code> should always be called
with an explicit condition type;
it should never simply be called with a string.
This enables internationalization.
</li>
<li>
Functions that report unusual outcomes
by signaling a condition should say so explicitly in their contracts
(their textual descriptions, in documentation and docstrings etc.).
When a function signals a condition
that is not specified by its contract, that's a bug.
The contract should specify the condition class(es) clearly.
The function may then signal any condition
that is a type-of any of those conditions.
That is, signaling instances of subclasses
of the documented condition classes is fine.
</li>
<li>
Complex bug-checks may need to use <code>ERROR</code>
instead of <code>ASSERT</code>.
</li>
<li>
When writing a server, you must not call <code>WARN</code>.
Instead, you should use the appropriate logging framework.
</li>
<li>
Code must not call <code>SIGNAL</code>.
Instead, use <code>ERROR</code> or <code>ASSERT</code>.
</li>
<li>
Code should not use <code>THROW</code> and <code>CATCH</code>;
instead use the <code>restart</code> facility.
</li>
<li>
Code should not generically handle all conditions,
e.g. type <code>T</code>, or use <code>IGNORE-ERRORS</code>.
Instead, let unknown conditions propagate to
the standard ultimate handler for processing.
</li>
<li>
There are a few places where handling all conditions is appropriate,
but they are rare.
The problem is that handling all conditions can mask program bugs.
If you <em>do</em> need to handle "all conditions",
you MUST handle only <code>ERROR</code>, <em>not</em> <code>T</code>
and not <code>SERIOUS-CONDITION</code>.
(This is notably because CCL's process shutdown
depends on being able to signal <code>process-reset</code>
and have it handled by CCL's handler,
so we must not interpose our own handler.)
</li>
<li>
<code>(error (make-condition 'foo-error ...))</code>
is equivalent to <code>(error 'foo-error ...)</code>
code must use the shorter form.
</li>
<li>
Code should not signal conditions from inside the cleanup form of
<code>UNWIND-PROTECT</code>
(unless they are always handled inside the cleanup form),
or otherwise do non-local exits from cleanup handlers
outside of the handler e.g. <code>INVOKE-RESTART</code>.
</li>
<li>
Do not clean up by resignaling.
If you do that, and the condition is not handled,
the stack trace will halt at the point of the resignal,
hiding the rest.
And the rest is the part we really care about!
<BAD_CODE_SNIPPET>
;; Bad
(handler-case
(catch 'ticket-at
(etd-process-blocks))
(error (c)
(reset-parser-values)
(error c)))
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
;; Better
(unwind-protect
(catch 'ticket-at
(etd-process-blocks))
(reset-parser-values))
</CODE_SNIPPET>
</li>
</ul>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Type Checking">
<SUMMARY>
If you know the type of something, you should make it explicit
in order to enable compile-time and run-time sanity-checking.
</SUMMARY>
<BODY>
<p>
If your function is using a special variable as an implicit argument,
it's good to put in a <code>CHECK-TYPE</code> for the special variable,
for two reasons:
to clue in the person reading the code
that this variable is being used implicitly as an argument,
and also to help detect bugs.
</p>
<p>
Using <code>(declare (type ...))</code>
is the least-desirable mechanism to use
because, as Scott McKay puts it:
</p>
<blockquote>
<p>
The fact is, <code>(declare (type ...))</code> does different things
depending on the compiler settings of speed, safety, etc.
In some compilers, when speed is greater than safety,
<code>(declare (type ...))</code> will tell the compiler
"please assume that these variables have these types"
<em>without</em> generating any type-checks.
That is, if some variable has the value <code>1432</code> in it,
and you declare it to be of type <code>string</code>,
the compiler might just go ahead and use it as though it's a string.
</p>
<p>
Moral: don't use <code>(declare (type ...))</code>
to declare the contract of any API functions,
it's not the right thing.
Sure, use it for "helper" functions, but not API functions.
</p>
</blockquote>
<p>
You should, of course, use appropriate declarations
in internal low-level functions
where these declarations are used for optimization.
When you do, however, see our recommendations for
<a href="#Unsafe_Operations">Unsafe Operations</a>.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="CLOS">
<SUMMARY>
Use CLOS appropriately.
</SUMMARY>
<BODY>
<p>
When a generic function is intended to be called from other
modules (other parts of the code), there should be an
explicit <code>DEFGENERIC</code> form,
with a <code>:DOCUMENTATION</code> string
explaining the generic contract of the function
(as opposed to its behavior for some specific class).
It's generally good to do explicit <code>DEFGENERIC</code> forms,
but for module entry points it is mandatory.
</p>
<p>
When the argument list of a generic function includes
<code>&amp;KEY</code>,
the <code>DEFGENERIC</code> should always explicitly list
all of the keyword arguments that are acceptable,
and explain what they mean.
(Common Lisp does not require this, but it is good form,
and it may avoid spurious warnings on SBCL.)
</p>
<p>
You should avoid <code>SLOT-VALUE</code> and <code>WITH-SLOTS</code>,
unless you absolutely intend to circumvent
any sort of method combination that might be in effect for the slot.
Rare exceptions include <code>INITIALIZE-INSTANCE</code>
and <code>PRINT-OBJECT</code> methods and
accessing normally hidden slots in the low-level implementation of
methods that provide user-visible abstractions.
Otherwise, you should use accessors,
<code>WITH-ACCESSORS</code>
</p>
<p>
Accessor names generally follow a convention of
<code>&lt;protocol-name&gt;-&lt;slot-name&gt;</code>,
where a "protocol" in this case loosely indicates
a set of functions with well-defined behavior.
</p>
<p>
No implication of a formal "protocol" concept is necessarily intended,
much less first-class "protocol" objects.
However, there may indeed be an abstract CLOS class
or an
<a href="https://common-lisp.net/~frideau/lil-ilc2012/lil-ilc2012.html">Interface-Passing Style</a> interface
that embodies the protocol.
Further (sub)classes or (sub)interfaces may then implement
all or part of a protocol by defining
some methods for (generic) functions in the protocol,
including readers and writers.
</p>
<p>
For example, if there were a notional protocol called
is <code>pnr</code> with accessors <code>pnr-segments</code>
and <code>pnr-passengers</code>, then
the classes <code>air-pnr</code>, <code>hotel-pnr</code> and
<code>car-pnr</code> could each reasonably implement
methods for <code>pnr-segments</code> and <code>pnr-passengers</code>
as accessors.
</p>
<p>
By default, an abstract base class name is used
as the notional protocol name, so accessor names default
to <code>&lt;class-name&gt;-&lt;slot-name&gt;</code>;
while such names are thus quite prevalent,
this form is neither required nor even preferred.
In general, it contributes to "symbol bloat",
and in many cases has led to a proliferation of "trampoline" methods.
</p>
<p>
Accessors named <code>&lt;slot-name&gt;-of</code> should not be used.
</p>
<p>
Explicit <code>DEFGENERIC</code> forms should be used when there are
(or it is anticipated that there will be)
more than one <code>DEFMETHOD</code> for that generic function.
The reason is that the documentation for the generic function
explains the abstract contract for the function,
as opposed to explaining what an individual method does for
some specific class(es).
</p>
<p>
You must not use generic functions where there is no notional protocol.
To put it more concretely,
if you have more than one generic function that specializes its Nth argument,
the specializing classes should all be descendants of a single class.
Generic functions must not be used for "overloading",
i.e. simply to use the same name for two entirely unrelated types.
</p>
<p>
More precisely, it's not really
whether they descend from a common superclass,
but whether they obey the same "protocol".
That is, the two classes should handle the same set of generic functions,
as if there were an explicit <code>DEFGENERIC</code> for each method.
</p>
<p>
Here's another way to put it.
Suppose you have two classes, A and B, and a generic function F.
There are two methods for F,
which dispatch on an argument being of types A and B.
Is it plausible that there might be a function call somewhere
in the program that calls F,
in which the argument might sometimes, at runtime,
be of class A and other times be of class B?
If not, you probably are overloading and
should not be using a single generic function.
</p>
<p>
We allow one exception to this rule:
it's OK to do overloading
if the corresponding argument "means" the same thing.
Typically one overloading allows an X object,
and the other allows the name of an X object,
which might be a symbol or something.
</p>
<p>
You must not use MOP "intercessory" operations at runtime.
You should not use MOP "intercessory" operations at compile-time.
At runtime, they are at worst a danger, at best a performance issue.
At compile-time, it is usually cleaner that
macros should set things up the right way in one pass
than have to require a second pass of fixups through intercession;
but sometimes, fixups are necessary to resolve forward references,
and intercession is allowed then.
MOP intercession is a great tool for interactive development,
and you may enjoy it while developping and debugging;
but you should not use it in normal applications.
</p>
<p>
If a class definition creates a method
as a <code>:READER</code>, <code>:WRITER</code>,
or <code>:ACCESSOR</code>,
do not redefine that method.
It's OK to add <code>:BEFORE</code>, <code>:AFTER</code>,
and <code>:AROUND</code> methods,
but don't override the primary method.
</p>
<p>
In methods with keyword arguments,
you must always use <code>&amp;KEY</code>,
even if the method does not care about the values of any keys,
and you should never use <code>&amp;ALLOW-OTHER-KEYS</code>.
As long as a keyword is accepted by any method of a generic function,
it's OK to use it in the generic function,
even if the other methods of the same generic function
don't mention it explicitly.
This is particularly important
for <code>INITIALIZE-INSTANCE</code> methods,
since if you did use <code>&amp;ALLOW-OTHER-KEYS</code>,
it would disable error checking for misspelled or wrong keywords
in <code>MAKE-INSTANCE</code> calls!
</p>
<p>
A typical <code>PRINT-OBJECT</code> method might look like this:
</p>
<CODE_SNIPPET>
(defmethod print-object ((p person) stream)
(print-unprintable-object (p stream :type t :identity t)
(with-slots (first-name last-name) p
(safe-format stream "~a ~a" first-name last-name))))
</CODE_SNIPPET>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Meta-language guidelines">
<STYLEPOINT title="Macros">
<SUMMARY>
Use macros when appropriate, which is often.
Define macros when appropriate, which is seldom.
</SUMMARY>
<BODY>
<p>
Macros bring syntactic abstraction, which is a wonderful thing.
It helps make your code clearer, by describing your intent
without getting bogged in implementation details
(indeed abstracting those details away).
It helps make your code more concise and more readable,
by eliminating both redundancy and irrelevant details.
But it comes at a cost to the reader,
which is learning a new syntactic concept for each macro.
And so it should not be abused.
</p>
<p>
The general conclusion is that there shouldn't be
any recognizable <em>design pattern</em>
in a good Common Lisp program.
The one and only pattern is: <em>use the language</em>,
which includes defining and using syntactic abstractions.
</p>
<p>
Existing macros must be used
whenever they make code clearer
by conveying intent in a more concise way,
which is often.
When a macro is available in your project
that expresses the concept you're using,
you must not write the expansion rather than use the macro.
</p>
<p>
New macros should be defined as appropriate,
which should be seldom,
for common macros have already been provided
by the language and its various libraries,
and your program typically only needs few new ones
relative to its size.
</p>
<p>
You should follow the OAOOM rule of thumb
for deciding when to create a new abstraction,
whether syntactic or not:
if a particular pattern is used more than twice,
it should probably be abstracted away.
A more refined rule to decide when to use abstraction
should take into account
the benefit in term of number of uses and gain at each use,
to the costs in term of having to get used to reading the code.
For syntactic abstractions, costs and benefits to the reader
is usually more important than costs and benefits to the writer,
because good code is usually written once
and read many times by many people
(including the same programmer
who has to maintain the program after having forgotten it).
Yet the cost to the writer of the macro
should also be taken into account;
however, in doing so it should rather be compared
to the cost of the programmer writing other code instead
that may have higher benefits.
</p>
<p>
Using Lisp macros properly requires taste.
Avoid writing complicated macros
unless the benefit clearly outweighs the cost.
It takes more effort for your fellow developers to learn your macro,
so you should only use a macro if the gain in expressiveness
is big enough to justify that cost.
As usual, feel free to consult your colleagues if you're not sure,
since without a lot of Lisp experience,
it can be hard to make this judgment.
</p>
<p>
You must never use a macro where a function will do.
That is, if the semantics of what you are writing
conforms to the semantics of a function,
then you must write it as a function rather than a macro.
</p>
<p>
You must not transform a function into a macro for performance reasons.
If profiling shows that you have a performance problem
with a specific function <code>FOO</code>,
document the need and profiling-results appropriately,
and
<code>(declaim (inline foo))</code>.
</p>
<p>
You can also use a compiler-macro
as a way to speed up function execution
by specifying a source-to-source transformation.
Beware that it interferes with tracing the optimized function.
</p>
<p>
When you write a macro-defining macro
(a macro that generates macros),
document and comment it particularly clearly,
since these are harder to understand.
</p>
<p>
You must not install new reader macros
without a consensus among the developers of your system.
Reader macros must not leak out of the system that uses them
to clients of that system or other systems used in the same project.
You must use software such as
<code>cl-syntax</code> or <code>named-readtables</code>
to control how reader macros are used.
This clients who desire it may use the same reader macros as you do.
In any case, your system must be usable
even to clients who do not use these reader macros.
</p>
<p>
If your macro has a parameter that is a Lisp form
that will be evaluated when the expanded code is run,
you should name the parameter with the suffix <code>-form</code>.
This convention helps make it clearer to the macro's user
which parameters are Lisp forms to be evaluated, and which are not.
The common names <code>body</code> and <code>end</code> are
exceptions to this rule.
</p>
<p>
You should follow the so-called <code>CALL-WITH</code> style when it applies.
This style is explained at length in
<a href="http://random-state.net/log/3390120648.html">http://random-state.net/log/3390120648.html</a>.
The general principle is that the macro is strictly limited to processing the syntax,
and as much of the semantics as possible is kept in normal functions.
Therefore, a macro <code>WITH-<em>FOO</em></code> is often limited to
generating a call to an auxiliary function
<code>CALL-WITH-<em>FOO</em></code>
with arguments deduced from the macro arguments.
Macro <code>&amp;body</code> arguments are typically
wrapped into a lambda expression of which they become the body,
which is passed as one of the arguments of the auxiliary function.
</p>
<p>
The separation of syntactic and semantic concerns
is a general principle of style that applies
beyond the case of <code>WITH-</code> macros.
Its advantages are many.
By keeping semantics outside the macro,
the macro is made simpler, easier to get right, and less subject to change,
which makes it easier to develop and maintain.
The semantics is written in a simpler language — one without staging —
which also makes it easier to develop and maintain.
It becomes possible to debug and update the semantic function
without having to recompile all clients of the macro.
The semantic function appears in the stack trace
which also helps debug client functions.
The macro expansion is made shorter and
each expansion shares more code with other expansions,
which reduces memory pressure which in turn usually makes things faster.
It also makes sense to write the semantic functions first,
and write the macros last as syntactic sugar on top.
You should use this style unless the macro is used
in tight loops where performance matters;
and even then, see our rules regarding optimization.
</p>
<p>
Any functions (closures) created by the macro should be named,
which can be done using <code>FLET</code>.
This also allows you to declare the function to be of dynamic extent
(if it is — and often it is; yet see below regarding
<a href="#DYNAMIC-EXTENT">DYNAMIC-EXTENT</a>).
</p>
<p>
If a macro call contains a form,
and the macro expansion includes more than one copy of that form,
the form can be evaluated more than once,
and code it contains macro-expanded and compiled more than once.
If someone uses the macro and calls it
with a form that has side effects or that takes a long time to compute,
the behavior will be undesirable
(unless you're intentionally writing
a control structure such as a loop).
A convenient way to avoid this problem
is to evaluate the form only once,
and bind a (generated) variable to the result.
There is a very useful macro called <code>ALEXANDRIA:ONCE-ONLY</code>
that generates code to do this.
See also <code>ALEXANDRIA:WITH-GENSYMS</code>,
to make some temporary variables in the generated code.
Note that if you follow our <code>CALL-WITH</code> style,
you typically expand the code only once, as either
an argument to the auxiliary function, or
the body of a lambda passed as argument to it;
you therefore avoid the above complexity.
</p>
<p>
When you write a macro with a body,
such as a <code>WITH-xxx</code> macro,
even if there aren't any parameters,
you should leave space for them anyway.
For example, if you invent <code>WITH-LIGHTS-ON</code>,
do not make the call to it look like
<code>(defmacro with-lights-on (&amp;body b) ...)</code>.
Instead, do <code>(defmacro with-lights-on (() &amp;body b) ...)</code>.
That way, if parameters are needed in the future,
you can add them without necessarily having to change
all the uses of the macro.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="EVAL-WHEN">
<SUMMARY>
When using <code>EVAL-WHEN</code>, you should almost always use all of
<code>(:compile-toplevel :load-toplevel :execute)</code>.
</SUMMARY>
<BODY>
<p>
Lisp evaluation happens at several times,
some of them interleaved.
Be aware of them when writing macros.
<a href="https://fare.livejournal.com/146698.html">EVAL-WHEN considered harmful to your mental health</a>.
</p>
<p>
In summary of the article linked above,
unless you're doing truly advanced macrology,
the only valid combination in an <code>EVAL-WHEN</code>
is to include all of
<code>(eval-when (:compile-toplevel :load-toplevel :execute) ...)</code>
</p>
<p>
You must use
<code>(eval-when (:compile-toplevel :load-toplevel :execute) ...)</code>
whenever you define functions, types, classes, constants, variables, etc.,
that are going to be used in macros.
</p>
<p>
It is usually an error to omit the <code>:execute</code>,
because it prevents <code>LOAD</code>ing the source rather than the fasl.
It is usually an error to omit the <code>:load-toplevel</code>
(except to modify e.g. readtables and compile-time settings),
because it prevents <code>LOAD</code>ing future files
or interactively compiling code
that depends on the effects that happen at compile-time,
unless the current file was <code>COMPILE-FILE</code>d
within the same Lisp session.
</p>
<p>
Regarding variables, note that because macros may or may not
be expanded in the same process that runs the expanded code,
you must not depend on compile-time and runtime effects
being either visible or invisible at the other time.
There are still valid uses of variables in macros:
</p>
<ul>
<li>
Some variables may hold dictionaries
for some new kind of definition and other meta-data.
If such meta-data is to be visible at runtime and/or in other files,
you must make sure that the macro expands into code that
will register the definitions to those meta-data structures
at load-time,
in addition to effecting the registration at compile-time.
Typically, your top-level definitions expand
to code that does the registration.
if your code doesn't expand at the top-level,
you can sometimes use <code>LOAD-TIME-VALUE</code> for good effect.
In extreme cases, you may have to use
<code>ASDF-FINALIZERS:EVAL-AT-TOPLEVEL</code>.
</li>
<li>
Some variables may hold temporary data
that is only used at compile-time in the same file,
and can be cleaned up at the end of the file's compilation.
Predefined such variables would include <code>*readtable*</code>
or compiler-internal variables holding
the current optimization settings.
You can often manage existing and new such variables using
the <code>:AROUND-COMPILE</code> hooks of <code>ASDF</code>.
</li>
</ul>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Read-time evaluation">
<SUMMARY>
You should use <code>#.</code> sparingly,
and you must avoid read-time side-effects.
</SUMMARY>
<BODY>
<p>
The <code>#.</code> standard read-macro
will read one object, evaluate the object, and
have the reader return the resulting value.
</p>
<p>
You must not use it where other idioms will do, such as
using <code>EVAL-WHEN</code> to evaluate side-effects at compile-time,
using a regular macro to return an expression computed at compile-time,
using <code>LOAD-TIME-VALUE</code> to compute it at load-time.
</p>
<p>
Read-time evaluation is often used as a quick way
to get something evaluated at compile time
(actually "read time" but it amounts to the same thing).
If you use this, the evaluation MUST NOT have any side effects
and MUST NOT depend on any variable global state.
The <code>#.</code> should be treated as a way
to force "constant-folding"
that a sufficiently-clever compiler
could have figure out all by itself,
when the compiler isn't sufficiently-clever
and the difference matters.
</p>
<p>
Another use of <code>#.</code> is to expand the equivalent of macros
in places that are neither expressions nor (quasi)quotations,
such as lambda-lists. However, if you find yourself using it a lot,
it might be time to instead define macros to replace your consumers
of lambda-lists with something that recognizes an extension.
</p>
<p>
Whenever you are going to use <code>#.</code>,
you should consider using <code>DEFCONSTANT</code> and its variants,
possibly in an <code>EVAL-WHEN</code>,
to give the value a name explaining what it means.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="EVAL">
<SUMMARY>
You must not use <code>EVAL</code> at runtime.
</SUMMARY>
<BODY>
<p>
Places where it is actually appropriate to use <code>EVAL</code>
are so few and far between that you must consult with your reviewers;
it's easily misused.
</p>
<p>
If your code manipulates symbols at runtime
and needs to get the value of a symbol,
use <code>SYMBOL-VALUE</code>, not <code>EVAL</code>.
</p>
<p>
Often, what you really need is to write a macro,
not to use <code>EVAL</code>.
</p>
<p>
You may be tempted to use <code>EVAL</code> as a shortcut
to evaluating expressions in a safe subset of the language.
But it often requires more scrutiny to properly check and sanitize
all possible inputs to such use of <code>EVAL</code>
than to build a special-purpose evaluator.
You must not use <code>EVAL</code> in this way at runtime.
</p>
<p>
Places where it is OK to use <code>EVAL</code> are:
</p>
<ul>
<li>
The implementation of an interactive development tool.
</li>
<li>
The build infrastructure.
</li>
<li>
Backdoors that are part of testing frameworks.
(You MUST NOT have such backdoors in production code.)
</li>
<li>
Macros that fold constants at compile-time.
</li>
<li>
Macros that register definitions to meta-data structures;
the registration form is sometimes evaluated at compile-time
as well as included in the macro-expansion,
so it is immediately available to other macros.
</li>
</ul>
<p>
Note that in the latter case,
if the macro isn't going to be used at the top-level,
it might not be possible to make these definitions available
as part of the expansion.
The same phenomenon may happen in a <code>DEFTYPE</code> expansion,
or in helper functions used by macros.
In these cases, you may actually have to use
<code>ASDF-FINALIZERS:EVAL-AT-TOPLEVEL</code> in your macro.
It will not only <code>EVAL</code> your definitions
at macro-expansion time for immediate availability,
it will also save the form aside, for inclusion in a
<code>(ASDF-FINALIZERS:FINAL-FORMS)</code>
that you need to include at the end of the file being compiled
(or before the form is needed).
This way, the side-effects are present when loading the fasl
without having compiled it as well as while compiling it;
in either case, the form is made available at load-time.
<code>ASDF-FINALIZERS</code> ensures that the form is present,
by throwing an error if you omit it.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="INTERN and UNINTERN">
<SUMMARY>
You must not use <code>INTERN</code> or <code>UNINTERN</code> at runtime.
</SUMMARY>
<BODY>
<p>
You must not use <code>INTERN</code> at runtime.
Not only does it cons,
it either creates a permanent symbol that won't be collected
or gives access to internal symbols.
This creates opportunities for memory leaks, denial of service attacks,
unauthorized access to internals, clashes with other symbols.
</p>
<p>
You must not <code>INTERN</code> a string
just to compare it to a keyword;
use <code>STRING=</code> or <code>STRING-EQUAL</code>.
</p>
<BAD_CODE_SNIPPET>
(member (intern str :keyword) $keys) ; Bad
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
(member str $keys :test #'string-equal) ; Better
</CODE_SNIPPET>
<p>
You must not use <code>UNINTERN</code> at runtime.
It can break code that relies on dynamic binding.
It makes things harder to debug.
You must not dynamically intern any new symbol,
and therefore you need not dynamically unintern anything.
</p>
<p>
You may of course use <code>INTERN</code> at compile-time,
in the implementation of some macros.
Even so, it is usually more appropriate
to use abstractions on top of it, such as
<code>ALEXANDRIA:SYMBOLICATE</code> or
<code>ALEXANDRIA:FORMAT-SYMBOL</code>
to create the symbols you need.
</p>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Data Representation">
<STYLEPOINT title="NIL: empty-list, false and I Don't Know">
<SUMMARY>
Appropriately use or avoid using <code>NIL</code>.
</SUMMARY>
<BODY>
<p>
<code>NIL</code> can have several different interpretations:
</p>
<ul>
<li>
"False."
In this case, use <code>NIL</code>.
You should test for false <code>NIL</code>
using the operator <code>NOT</code> or
using the predicate function <code>NULL</code>.
</li>
<li>
"Empty-list."
In this case, use <code>'()</code>.
(Be careful about quoting the empty-list when calling macros.)
You should use <code>ENDP</code> to test for the empty list
when the argument is known to be a proper list,
or with <code>NULL</code> otherwise.
</li>
<li>
A statement about some value being unspecified.
In this case, you may use <code>NIL</code>
if there is no risk of ambiguity anywhere in your code;
otherwise you should use an explicit, descriptive symbol.
</li>
<li>
A statement about some value being known not to exist.
In this case, you should use an explicit, descriptive symbol
instead of <code>NIL</code>.
</li>
</ul>
<p>
You must not introduce ambiguity in your data representations
that will cause headaches for whoever has to debug code.
If there is any risk of ambiguity,
you should use an explicit, descriptive symbol or keyword
for each case,
instead of using <code>NIL</code> for either.
If you do use <code>NIL</code>,
you must make sure that the distinction is well documented.
</p>
<p>
In many contexts,
instead of representing "I don't know" as a particular value,
you should instead use multiple values,
one for the value that is known if any,
and one to denote whether the value was known or found.
</p>
<p>
When working with database classes, keep in mind that
<code>NIL</code> need not always map to <code>'NULL'</code>
(and vice-versa)!
The needs of the database may differ from the needs of the Lisp.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Do not abuse lists">
<SUMMARY>
You must select proper data representation.
You must not abuse the <code>LIST</code> data structure.
</SUMMARY>
<BODY>
<p>
Even though back in 1958, LISP was short for "LISt Processing",
its successor Common Lisp has been a modern programming language
with modern data structures since the 1980s.
You must use the proper data structures in your programs.
</p>
<p>
You must not abuse the builtin (single-linked) <code>LIST</code>
data structure where it is not appropriate,
even though Common Lisp makes it especially easy to use it.
</p>
<p>
You must only use lists
when their performance characteristics
is appropriate for the algorithm at hand:
sequential iteration over the entire contents of the list.
</p>
<p>
An exception where it is appropriate to use lists
is when it is known in advance
that the size of the list will remain very short
(say, less than 16 elements).
</p>
<p>
List data structures are often (but not always)
appropriate for macros and functions used by macros at compile-time:
indeed, not only is source code passed as lists in Common Lisp,
but the macro-expansion and compilation processes
will typically walk over the entire source code, sequentially, once.
(Note that advanced macro systems don't directly use lists, but instead
use abstract syntax objects that track source code location and scope;
however there is no such advanced macro system
in Common Lisp at this time.)
</p>
<p>
Another exception where it is appropriate to use lists is
for introducing literal constants
that will be transformed into more appropriate data structures
at compile-time or load-time.
It is a good to have a function with a relatively short name
to build your program's data structures from such literals.
</p>
<p>
In the many cases when lists are not the appropriate data structure,
various libraries such as
<a href="http://cliki.net/cl-containers">cl-containers</a> or
<a href="http://cliki.net/lisp-interface-library">lisp-interface-library</a>
provide plenty of different data structures
that should fulfill all the basic needs of your programs.
If the existing libraries are not satisfactory, see above about
<a href="#Using_Libraries">Using Libraries</a> and
<a href="#Open-Sourcing_Code">Open-Sourcing Code</a>.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Lists vs. structures vs. multiple values">
<SUMMARY>
You should use the appropriate representation for product types.
</SUMMARY>
<BODY>
<p>
You should avoid using a list as anything
besides a container of elements of like type.
You must not use a list as method of passing
multiple separate values of different types
in and out of function calls.
Sometimes it is convenient to use a list
as a little ad hoc structure,
i.e. "the first element of the list is a FOO, and the second is a BAR",
but this should be used minimally
since it gets harder to remember the little convention.
You must only use a list that way
when destructuring the list of arguments from a function,
or creating a list of arguments
to which to <code>APPLY</code> a function.
</p>
<p>
The proper way to pass around an object
comprising several values of heterogeneous types
is to use a structure as defined by <code>DEFSTRUCT</code>
or <code>DEFCLASS</code>.
</p>
<p>
You should use multiple values only
when function returns a small number of values
that are meant to be destructured immediately by the caller,
rather than passed together as arguments to further functions.
</p>
<p>
You should not return a condition object
as one of a set of multiple values.
Instead, you should signal the condition to denote an unusual outcome.
</p>
<p>
You should signal a condition to denote an unusual outcome,
rather than relying on a special return type.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Lists vs. Pairs">
<SUMMARY>
Use the appropriate functions when manipulating lists.
</SUMMARY>
<BODY>
<p>
Use <code>FIRST</code> to access the first element of a list,
<code>SECOND</code> to access the second element, etc.
Use <code>REST</code> to access the tail of a list.
Use <code>ENDP</code> to test for the end of the list.
</p>
<p>
Use <code>CAR</code> and <code>CDR</code>
when the cons cell is not being used to implement a proper list
and is instead being treated as a pair of more general objects.
Use <code>NULL</code> to test for <code>NIL</code> in this context.
</p>
<p>
The latter case should be rare outside of alists,
since you should be using structures and classes where they apply,
and data structure libraries when you want trees.
</p>
<p>
Exceptionally, you may use <code>CDADR</code> and other variants
on lists when manually destructuring them,
instead of using a combination of several list accessor functions.
In this context, using <code>CAR</code> and <code>CDR</code>
instead of <code>FIRST</code> and <code>REST</code> also makes sense.
However, keep in mind that it might be more appropriate in such cases
to use higher-level constructs such as
<code>DESTRUCTURING-BIND</code> or <code>OPTIMA:MATCH</code>.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Lists vs. Arrays">
<SUMMARY>
You should use arrays rather than lists where random access matters.
</SUMMARY>
<BODY>
<p>
<code>ELT</code> has <i>O(n)</i> behavior when used on lists.
If you are to use random element access on an object,
use arrays and <code>AREF</code> instead.
</p>
<p>
The exception is for code outside the critical path
where the list is known to be small anyway.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Lists vs. Sets">
<SUMMARY>
You should only use lists as sets for very small lists.
</SUMMARY>
<BODY>
<p>
Using lists as representations of sets is a bad idea
unless you know the lists will be small,
for accessors are <i>O(n)</i> instead of <i>O(log n)</i>.
For arbitrary big sets, use balanced binary trees,
for instance using <code>lisp-interface-library</code>.
</p>
<p>
If you still use lists as sets,
you should not <code>UNION</code> lists just to search them.
</p>
<BAD_CODE_SNIPPET>
(member foo (union list-1 list-2)) ; Bad
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
(or (member foo list-1) (member foo list-2)) ; Better
</CODE_SNIPPET>
<p>
Indeed, <code>UNION</code> not only conses unnecessarily,
but it can be <i>O(n^2)</i> on some implementations,
and is rather slow even when it's <i>O(n)</i>.
</p>
</BODY>
</STYLEPOINT>
</CATEGORY>
<CATEGORY title="Proper Forms">
<p>
You must follow the proper usage regarding
well-known functions, macros and special forms.
</p>
<STYLEPOINT title="Defining Constants">
<SUMMARY>
You must use proper defining forms for constant values.
</SUMMARY>
<BODY>
<p>
The Lisp system we primarily use, SBCL, is very picky and
signals a condition whenever a constant is redefined to a value not
<code>EQL</code> to its previous setting.
You must not use <code>DEFCONSTANT</code>
when defining variables that are not
numbers, characters, or symbols (including booleans and keywords).
Instead, consistently use whichever alternative
is recommended for your project.
</p>
<BAD_CODE_SNIPPET>
;; Bad
(defconstant +google-url+ "https://www.google.com/")
(defconstant +valid-colors+ '(red green blue))
</BAD_CODE_SNIPPET>
<p>
Open-Source libraries may use
<code>ALEXANDRIA:DEFINE-CONSTANT</code>
for constants other than numbers, characters and symbols
(including booleans and keywords).
You may use the <code>:TEST</code> keyword argument
to specify an equality predicate.
</p>
<CODE_SNIPPET>
;; Better, for Open-Source code:
(define-constant +google-url+ "https://www.google.com/" :test #'string=)
(define-constant +valid-colors+ '(red green blue))
</CODE_SNIPPET>
<p>
Note that with optimizing implementations, such as SBCL or CMUCL,
defining constants this way precludes any later redefinition
short of <code>UNINTERN</code>ing the symbol
and recompiling all its clients.
This may make it "interesting" to debug things at the REPL
or to deploy live code upgrades.
If there is a chance that your "constants" are not going to be constant
over the lifetime of your server processes
after taking into consideration scheduled and unscheduled code patches,
you should consider using
<code>DEFPARAMETER</code> or <code>DEFVAR</code> instead,
or possibly a variant of <code>DEFINE-CONSTANT</code>
that builds upon some future library implementing global lexicals
rather than <code>DEFCONSTANT</code>.
You may keep the <code>+plus+</code> convention in these cases
to document the intent of the parameter as a constant.
</p>
<p>
Also note that <code>LOAD-TIME-VALUE</code> may help you
avoid the need for defined constants.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Defining Functions">
<SUMMARY>
You should make proper use of
<code>&amp;OPTIONAL</code> and
<code>&amp;KEY</code> arguments.
You should not use <code>&amp;AUX</code> arguments.
</SUMMARY>
<BODY>
<p>
You should avoid using <code>&amp;ALLOW-OTHER-KEYS</code>,
since it blurs the contract of a function.
Almost any real function (generic or not) allows a certain
fixed set of keywords, as far as its caller is concerned,
and those are part of its contract.
If you are implementing a method of a generic function,
and it does not need to know
the values of some of the keyword arguments,
you should explicitly <code>(DECLARE (IGNORE ...))</code>
all the arguments that you are not using.
You must not use <code>&amp;ALLOW-OTHER-KEYS</code>
unless you explicitly want to disable checking of allowed keys
for all methods when invoking the generic function on arguments
that match this particular method.
Note that the contract of a generic function belongs in
the <code>DEFGENERIC</code>, not in the <code>DEFMETHOD</code>
which is basically an "implementation detail" of the generic function
as far as the caller of the generic is concerned.
</p>
<p>
A case where <code>&amp;ALLOW-OTHER-KEYS</code> is appropriate
is when you write a wrapper function to other some other functions
that may vary (within the computation or during development),
and pass around a plist as a <code>&amp;REST</code> argument.
</p>
<p>
You should avoid using <code>&amp;AUX</code> arguments.
</p>
<p>
You should avoid having both <code>&amp;OPTIONAL</code>
and <code>&amp;KEY</code> arguments,
unless it never makes sense to specify keyword arguments
when the optional arguments are not all specified.
You must not have non-<code>NIL</code> defaults
to your <code>&amp;OPTIONAL</code> arguments
when your function has both <code>&amp;OPTIONAL</code>
and <code>&amp;KEY</code> arguments.
</p>
<p>
For maximum portability of a library, it is good form
that <code>DEFMETHOD</code> definitions should
<code>(DECLARE (IGNORABLE ...))</code>
all the required arguments that they are not using.
Indeed, some implementations will issue a warning
if you <code>(DECLARE (IGNORE ...))</code> those arguments,
whereas other implementations will issue a warning
if you fail to <code>(DECLARE (IGNORE ...))</code> them.
<code>(DECLARE (IGNORABLE ...))</code> works on all implementations.
</p>
<p>
You should avoid excessive nesting of binding forms inside a function.
If your function ends up with massive nesting,
you should probably break it up into several functions or macros.
If it is really a single conceptual unit,
consider using a macro such as <code>FARE-UTILS:NEST</code>
to at least reduce the amount of indentation required.
It is bad form to use <code>NEST</code> in typical short functions
with 4 or fewer levels of nesting,
but also bad form not to use it in the exceptional long functions
with 10 or more levels of nesting.
Use your judgment and consult your reviewers.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Conditional Expressions">
<SUMMARY>
Use the appropriate conditional form.
</SUMMARY>
<BODY>
<p>
Use <code>WHEN</code> and <code>UNLESS</code>
when there is only one alternative.
Use <code>IF</code> when there are two alternatives
and <code>COND</code> when there are several.
</p>
<p>
However, don't use <code>PROGN</code> for an <code>IF</code> clause
— use <code>COND</code>, <code>WHEN</code>, or <code>UNLESS</code>.
</p>
<p>
Note that in Common Lisp,
<code>WHEN</code> and <code>UNLESS</code> return <code>NIL</code>
when the condition is not met.
You may take advantage of it.
Nevertheless, you may use an <code>IF</code>
to explicitly return <code>NIL</code>
if you have a specific reason to insist on the return value.
You may similarly include a fall-through clause <code>(t nil)</code>
as the last in your <cond>COND</cond>,
or <code>(otherwise nil)</code> as the last in your <cond>CASE</cond>,
to insist on the fact that the value returned by the conditional matters
and that such a case is going to be used.
You should omit the fall-through clause
when the conditional is used for side-effects.
</p>
<p>
You should prefer <code>AND</code> and <code>OR</code>
when it leads to more concise code than using
<code>IF</code>, <code>COND</code>,
<code>WHEN</code> or <code>UNLESS</code>,
and there are no side-effects involved.
You may also use an <code>ERROR</code>
as a side-effect in the final clause of an <code>OR</code>.
</p>
<p>
You should only use <code>CASE</code> and <code>ECASE</code>
to compare numbers, characters or symbols
(including booleans and keywords).
Indeed, <code>CASE</code> uses <code>EQL</code> for comparisons,
so strings, pathnames and structures may not compare the way you expect,
and <code>1</code> will differ from <code>1.0</code>.
</p>
<p>
You should use <code>ECASE</code> and <code>ETYPECASE</code>
in preference to <code>CASE</code> and <code>TYPECASE</code>.
It is better to catch erroneous values early.
</p>
<p>
You should not use <code>CCASE</code> or <code>CTYPECASE</code> at all.
At least, you should not use them in server processes,
unless you have quite robust error handling infrastructure
and make sure not to leak sensitive data this way.
These are meant for interactive use,
and can cause interesting damage
if they cause data or control to leak to attackers.
</p>
<p>
You must not use gratuitous single quotes in <code>CASE</code> forms.
This is a common error:
</p>
<BAD_CODE_SNIPPET>
(case x ; Bad: silently returns NIL on mismatch
('bar :bar) ; Bad: catches QUOTE
('baz :baz)) ; Bad: also would catch QUOTE
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
(ecase x ; Better: will error on mismatch
((bar) :bar) ; Better: won't match QUOTE
((baz) :baz)) ; Better: same reason
</CODE_SNIPPET>
<p>
<code>'BAR</code> there is <code>(QUOTE BAR)</code>,
meaning this leg of the case will be executed
if <code>X</code> is <code>QUOTE</code>...
and ditto for the second leg
(though <code>QUOTE</code> will be caught by the first clause).
This is unlikely to be what you really want.
</p>
<p>
In <code>CASE</code> forms,
you must use <code>otherwise</code> instead of <code>t</code>
when you mean "execute this clause if the others fail".
You must use <code>((t) ...)</code>
when you mean "match the symbol T" rather than "match anything".
You must also use <code>((nil) ...)</code>
when you mean "match the symbol NIL" rather than "match nothing".
</p>
<p>
Therefore, if you want to map booleans <code>NIL</code> and <code>T</code>
to respective symbols <code>:BAR</code> and <code>:QUUX</code>,
you should avoid the former way and do it the latter way:
</p>
<BAD_CODE_SNIPPET>
(ecase x ; Bad: has no actual error case!
(nil :bar)) ; Bad: matches nothing
(t :quux)) ; Bad: matches anything
</BAD_CODE_SNIPPET>
<CODE_SNIPPET>
(ecase x ; Better: will actually catch non-booleans
((nil) :bar)) ; Better: matches NIL
((t) :quux)) ; Better: matches T
</CODE_SNIPPET>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Identity, Equality and Comparisons">
<SUMMARY>
You should use the appropriate predicates when comparing objects.
</SUMMARY>
<BODY>
<p>
Lisp provides four general equality predicates:
<code>EQ</code>, <code>EQL</code>, <code>EQUAL</code>,
and <code>EQUALP</code>,
which subtly vary in semantics.
Additionally, Lisp provides the type-specific predicates
<code>=</code>, <code>CHAR=</code>, <code>CHAR-EQUAL</code>,
<code>STRING=</code>, and <code>STRING-EQUAL</code>.
Know the distinction!
</p>
<p>
You should use <code>EQL</code> to compare objects and symbols
for <em>identity</em>.
</p>
<p>
You must not use <code>EQ</code> to compare numbers or characters.
Two numbers or characters that are <code>EQL</code>
are not required by Common Lisp to be <code>EQ</code>.
</p>
<p>
When choosing between <code>EQ</code> and <code>EQL</code>,
you should use <code>EQL</code> unless you are writing
performance-critical low-level code.
<code>EQL</code> reduces the opportunity
for a class of embarrassing errors
(i.e. if numbers or characters are ever compared).
There may a tiny performance cost relative to <code>EQ</code>,
although under SBCL, it often compiles away entirely.
<code>EQ</code> is equivalent to <code>EQL</code> and type declarations,
and use of it for optimization should be treated just like
any such <a href="#Unsafe_Operations">unsafe operations</a>.
</p>
<p>
You should use <code>CHAR=</code>
for case-dependent character comparisons,
and <code>CHAR-EQUAL</code> for case-ignoring character comparisons.
</p>
<p>
You should use <code>STRING=</code>
for case-dependent string comparisons,
and <code>STRING-EQUAL</code> for case-ignoring string comparisons.
</p>
<p>
A common mistake when using <code>SEARCH</code> on strings
is to provide <code>STRING=</code> or <code>STRING-EQUAL</code>
as the <code>:TEST</code> function.
The <code>:TEST</code> function
is given two sequence elements to compare.
If the sequences are strings,
the <code>:TEST</code> function is called on two characters,
so the correct tests are <code>CHAR=</code> or <code>CHAR-EQUAL</code>.
If you use <code>STRING=</code> or <code>STRING-EQUAL</code>,
the result is what you expect,
but in some Lisp implementations it's much slower.
CCL (at least as of 8/2008)
creates a one-character string upon each comparison, for example,
which is very expensive.
</p>
<p>
Also, you should use <code>:START</code> and <code>:END</code> arguments
to <code>STRING=</code> or <code>STRING-EQUAL</code>
instead of using <code>SUBSEQ</code>;
e.g. <code>(string-equal (subseq s1 2 6) s2)</code> should instead be
<code>(string-equal s1 s2 :start1 2 :end1 6)</code>
This is preferable because it does not cons.
</p>
<p>
You should use <code>ZEROP</code>,
<code>PLUSP</code>, or <code>MINUSP</code>,
instead of comparing a value to <code>0</code> or <code>0.0</code>.
</p>
<p>
You must not use exact comparison on floating point numbers,
since the vague nature of floating point arithmetic
can produce little "errors" in numeric value.
You should compare absolute values to a threshhold.
</p>
<p>
You must use <code>=</code> to compare numbers,
unless you really mean for <code>0</code>,
<code>0.0</code> and <code>-0.0</code> to compare unequal,
in which case you should use <code>EQL</code>.
Then again, you must not usually use exact comparison
on floating point numbers.
</p>
<p>
Monetary amounts should be using decimal (rational) numbers
to avoid the complexities and rounding errors
of floating-point arithmetic.
Libraries such as
<a href="http://wukix.com/lisp-decimals">wu-decimal</a>
may help you;
once again, if this library is not satisfactory, see above about
<a href="#Using_Libraries">Using Libraries</a> and
<a href="#Open-Sourcing_Code">Open-Sourcing Code</a>.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="Iteration">
<SUMMARY>
Use the appropriate form for iteration.
</SUMMARY>
<BODY>
<p>
You should use simpler forms such as
<code>DOLIST</code> or <code>DOTIMES</code>
instead of <code>LOOP</code>
in simple cases when you're not going to use any
of the <code>LOOP</code> facilities such as
bindings, collection or block return.
</p>
<p>
Use the <code>WITH</code> clause of <code>LOOP</code>
when it will avoid a level of nesting with <code>LET</code>.
You may use <code>LET</code> if it makes it clearer
to return one of bound variables after the <code>LOOP</code>,
rather than use a clumsy <code>FINALLY (RETURN ...)</code> form.
</p>
<p>
In the body of a <code>DOTIMES</code>,
do not set the iteration variable.
(CCL will issue a compiler warning if you do.)
</p>
<p>
Most systems use unadorned symbols in the current package
as <code>LOOP</code> keywords.
Other systems use actual <code>:keywords</code>
from the <code>KEYWORD</code> package
as <code>LOOP</code> keywords.
You must be consistent with the convention used in your system.
</p>
</BODY>
</STYLEPOINT>
<STYLEPOINT title="I/O">
<SUMMARY>
Use the appropriate I/O functions.
</SUMMARY>
<BODY>
<p>
When writing a server,
code must not send output to the standard streams such as
<code>*STANDARD-OUTPUT*</code> or <code>*ERROR-OUTPUT*</code>.
Instead, code must use the proper logging framework
to output messages for debugging.
We are running as a server, so there is no console!
</p>
<p>
Code must not use <code>PRINT-OBJECT</code>
to communicate with a user —
<code>PRINT-OBJECT</code> is for debugging purposes only.
Modifying any <code>PRINT-OBJECT</code> method
must not break any public interfaces.
</p>
<p>
You should not use a sequence of <code>WRITE-XXX</code>
where a single <code>FORMAT</code> string could be used.
Using format allows you
to parameterize the format control string in the future
if the need arises.
</p>
<p>
You should use <code>WRITE-CHAR</code> to emit a character
rather than <code>WRITE-STRING</code>
to emit a single-character string.
</p>
<p>
You should not use <code>(format nil "~A" value)</code>;
you should use <code>PRINC-TO-STRING</code> instead.
</p>
<p>
You should use <code>~&lt;Newline&gt;</code>
or <code>~@&lt;Newline&gt;</code> in format strings
to keep them from wrapping in 100-column editor windows,
or to indent sections or clauses to make them more readable.
</p>
<p>
You should not use <code>STRING-UPCASE</code>
or <code>STRING-DOWNCASE</code>
on format control parameters;
instead, it should use <code>"~:@(~A~)"</code> or <code>"~(~A~)"</code>.
</p>
<p>
Be careful when using the <code>FORMAT</code> conditional directive.
The parameters are easy to forget.
</p>
<dl>
<dt>No parameters, e.g. <code>"~[Siamese~;Manx~;Persian~] Cat"</code></dt>
<dd>
Take one format argument, which should be an integer.
Use it to choose a clause. Clause numbers are zero-based.
If the number is out of range, just print nothing.
You can provide a default value
by putting a <code>":"</code> in front of the last <code>";"</code>.
E.g. in <code>"~[Siamese~;Manx~;Persian~:;Alley~] Cat"</code>,
an out-of-range arg prints <code>"Alley"</code>.
</dd>