-
Notifications
You must be signed in to change notification settings - Fork 5
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
self code review #243
Labels
Comments
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
chrisklus
added a commit
that referenced
this issue
Jul 31, 2019
Self review complete. I'll create an issue for the real review. |
Closed
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
As per discussion with @ariel-phet and @jbphet on 07/29/19, the formal code review is just going to start with EFACIntroModel.js, EFACIntroScreenView.js, SystemsModel.js, and SystemsScreenView.js. I'm going self review these before opening an issue for the final review.
Build and Run Checks
Does the sim build without warnings or errors?
Does the html file size seem reasonable, compared to other similar sims?
Does the sim start up? (requirejs and built versions)
Does the sim experience any assertion failures? (run with query parameter
ea
)Does the sim pass a scenery fuzz test? (run with query parameters
fuzz&ea
)Does linting with "sim_es6_eslintrc_review.js" reveal any problems that should be fixed? (change eslintConfig in package.json and run
grunt lint
)UPDATE: Lots of errors, but mostly lines exceeding 120 chars for other parts of the sim. I checked out a few of the other errors and was not concerned.
Memory Leaks
Does a heap comparison using Chrome Developer Tools indicate a memory leak? (Describing this process is beyond the scope of this document.) Test on a version built using
grunt --minify.mangle=false
. There should be a GitHub issue showing the results of testing done by the primary developer.For each common-code component (sun, scenery-phet, vegas, …) that opaquely registers observers or listeners, is
there a call to that component’s
dispose
function, or is it obvious why it isn't necessary, or is there documentationabout why
dispose
isn't called? An example of why no call todispose
is needed is if the component is used ina screen view that would never be removed from the scene graph.
Are there leaks due to registering observers or listeners? The following guidelines should be followed unless
there it is obviously no need to unlink, or documentation (in-line or in the implementation nodes)added about why
following them is not necessary. Examples of when an unlink is not needed is if the property never
Property.link
is accompanied byProperty.unlink
.DerivedProperty
is accompanied bydispose
.Multilink
is accompanied bydispose
.Events.on
is accompanied byEvents.off
.Emitter.addListener
is accompanied byEmitter.removeListener
.Node.on
is accompanied byNode.off
PhetioObject
instances should be disposed.Do all types that require a
dispose
function have one? This should expose a publicdispose
function that callsthis.disposeMyType()
, wheredisposeMyType
is a private function declared in the constructor.MyType
should exactly match the filename.PhET-iO instantiates different objects and wires up listeners that are not present in the PhET-branded simulation. It needs to be tested separately for memory leaks. To help isolate the nature of the memory leak, this test should be run separately from the PhET brand memory leak test. Test with the "console" and "studio" wrappers (easily accessed from phetmarks)
UPDATE: Memory will be tested again over in Memory test before 1.2 #245.
Performance
webgl=false
)Usability
Are UI components sufficiently responsive? (especially continuous UI components, such as sliders)
Are pointer areas optimized, especially for touch? (run with query parameter
showPointerAreas
)UPDATE: Nearly complete, remaining work in Usability on small screens review #239
Do pointer areas overlap? (run with query parameter
showPointerAreas
) Some overlap may be OK depending on the z-ordering (if the frontmost object is supposed to occlude touch/mouse areas)Internationalization
Are there any strings that are not being internationalized? (run with query parameter
stringTest=x
, you should see nothing but 'x' strings)Does the sim layout gracefully handle internationalized strings that are twice as long as the English strings? (run with query parameter
stringTest=double
)Does the sim layout gracefully handle internationalized strings that are exceptionally long? (run with query parameter
stringTest=long
)Does the sim layout gracefully handle internationalized strings that are shorter than the English strings? (run with query parameter
stringTest=X
)Does the sim stay on the sim page (doesn't redirect to an external page) when running with the query parameter
stringTest=xss
? This test passes if sim does not redirect, OK if sim crashes or fails to fully start. Only test on onedesktop platform. For PhET-iO sims, additionally test
?stringTest=xss
in Studio to make sure i18n strings didn't leakto phetioDocumentation, see https://github.com/phetsims/phet-io/issues/1377
Use named placeholders (e.g.
"{{value}} {{units}}"
) instead of numbered placeholders (e.g."{0} {1}"
).Make sure the string keys are all perfect - they are difficult to change after 1.0.0 is published. Guidelines for string keys are:
(1) Strings keys should generally match their values. E.g.:
(2) If a string key would be exceptionally long, use a key name that is an abbreviated form of the string value, or that captures the purpose/essence of the value. E.g.:
(3) If string key names would collide, use your judgment to disambiguate. E.g.:
(4) String keys for screen names should have the general form
"screen.{{screenName}}"
. E.g.:(5) String patterns that contain placeholders (e.g.
"My name is {{first}} {{last}}"
) should use keys that are unlikely to conflict with strings that might be needed in the future. For example, for"{{price}}"
consider using key"pricePattern"
instead of"price"
, if you think there might be a future need for a"price"
string.Repository Structure
Are all required files and directories present?
For a sim repository named “my-repo”, the general structure should look like this (where assets/, sound/ or images/ may be omitted if the sim doesn’t have those types of assets).
*Any images used in model.md or implementation-notes.md should be added here. Images specific to aiding with documentation do not need their own license.
Is the js/ directory properly structured?
All JavaScript source should be in the js/ directory. There should be a subdirectory for each screen (this also applies for single-screen sims, where the subdirectory matches the repo name). For a multi-screen sim, code shared by 2 or more screens should be in a js/common/ subdirectory. Model and view code should be in model/ and view/ subdirectories for each screen and common/. For example, for a sim with screens “Introduction” and “Lab”, the general directory structure should look like this:
Do filenames use an appropriate prefix? Some filenames may be prefixed with the repository name, e.g.
MolarityConstants.js
in molarity. If the repository name is long, the developer may choose to abbreviate the repository name, e.g.EEConstants.js
in expression-exchange. If the abbreviation is already used by another respository, then the full name must be used. For example, if the "EE" abbreviation is already used by expression-exchange, then it should not be used in equality-explorer. Whichever convention is used, it should be used consistently within a repository - don't mix abbreviations and full names.Is there a file in assets/ for every resource file in sound/ and images/? Note that there is not necessarily a 1:1 correspondence between asset and resource files; for example, several related images may be in the same .ai file. Check license.json for possible documentation of why some reesources might not have a corresponding asset file.
UPDATE: Nope, and I don't think this could be done easily or is worth the design time. Some of the images used are pretty old and are without .ai files.
Was the README.md generated by
grunt published-README
orgrunt unpublished-README
?Does package.json refer to any dependencies that are not used by the sim?
UPDATE: Issue for removing unneeded dependencies: Remove unneeded dependencies from dependencies.json #246
Is the sim's -config.js up-to-date (generated by
grunt generate-config
)Is the LICENSE file correct? (Generally GPL v3 for sims and MIT for common code, see this thread for additional information).
Does .gitignore match the one in simula-rasa?
Does a GitHub issue exist for tracking credits, to ensure that they are correct before publication?
In GitHub, verify that all non-release branches have an associated issue that describes their purpose.
Are there any GitHub branches that are no longer needed and should be deleted?
Does
model.md
adequately describe the model, in terms appropriate for teachers?UPDATE: see Update implementation-notes.md and model.md #244
Does
implementation-notes.md
adequately describe the implementation, with an overview that will be useful to future maintainers?UPDATE: see Update implementation-notes.md and model.md #244
Are sim-specific query parameters (if any) identified and documented in one .js file in js/common/ or js/ (if there is no common/)? The .js file should be named
{{REPO}}QueryParameters.js
, for example ArithmeticQueryParameters.js for the aritmetic repository.Coding Conventions
Is the code formatted according to PhET conventions? See phet-idea-code-style.xml for IntelliJ IDEA code style.
Are copyright headers present and up to date? Run
grunt update-copyright-dates
.Names (types, variables, properties, functions,...) should be sufficiently descriptive and specific, and should avoid non-standard abbreviations. For example:
Require statements should be organized into blocks, with the code modules first, followed by plugins (strings, images, sound, ifphetio - any order ok for plugins). For modules, the variable name should match the file name. Example below.
Do the
@author
annotations seem correct?Are all constructors marked with
@constructor
? That will make them easier to search and review. This is not necessary for ES6 constructors.For constructors, use parameters for things that don’t have a default. Use options for things that have a default value. This improves readability at the call site, especially when the number of parameters is large. It also eliminates order dependency that is required by using parameters.
For example, this constructor uses parameters for everything. At the call site, the semantics of the arguments are difficult to determine without consulting the constructor.
Here’s the same constructor with an appropriate use of options. The call site is easier to read, and the order of options is flexible.
When options are passed through one constructor to another, a "nested options" pattern should be used. This helps to avoid duplicating option names and/or accidentally overwriting options for different components that use the same option names.
Example:
A possible exception to this guideline is when the constructor API is improved by hiding the implementation details, i.e. not revealing that a sub-component exists. In that case, it may make sense to use new top-level options. This is left to developer and reviewer discretion.
For more information on the history and thought process around the "nested options" pattern, please see A better way to pass through configuration tasks#730.
Constructor and function documentation. Parameter types and names should be clearly specified for each constructor and function using
@param
annotations. The description for each parameter should follow a hyphen. Primitive types should use lower case. For example:For most functions, the same form as above should be used, with a
@returns
annotation which identifies the return type and the meaning of the returned value. Functions should also document any side effects. For extremely simple functions that are just a few lines of simple code, an abbreviated line-comment can be used, for example:// Computes {Number} distance based on {Foo} foo.
If references are needed to the enclosing object, such as for a closure,
self
should be defined, but it should only be used in closures. Theself
variable should not be defined unless it is needed in a closure. Example:Generally, lines should not exceed 120 columns. Break up long statements, expressions, or comments into multiple
lines to optimize readability. It is OK for require statements or other structured patterns to exceed 120 columns.
Use your judgment!
Use
class
andextends
for defining classes and implementing inheritance.PHET_CORE/inherit
was a pre-ES6 implementation of inheritance that is specific to PhET and has been supplanted byclass
andextends
.inherit
shouldnot be used in new code.
Functions should be invoked using the dot operator rather than the bracket operator. For more details, please see Style question: bracket notation vs dot notation gravity-and-orbits#9. For example:
It is not uncommon to use conditional shorthand and short circuiting for invocation. Use parentheses to maximize readability.
If the expression is only one item, the parentheses can be omitted. This is the most common use case.
Naming for Property values: All
AXON/Property
instances should be declared with the suffixProperty
. For example, if a visible property is added, it should have the namevisibleProperty
instead of simplyvisible
. This will help to avoid confusion with non-Property definitions.Properties should use type-specific subclasses where appropriate (.e.g BooleanProperty, NumberProperty, StringProperty) or provide documentation as to why they are not.
Are Property value validation options (
valueType
,validValues
, etc...) utilized? Is their presence or lack thereof properly documented?Line comments should generally be preceded by a blank line. For example:
Line comments should have whitespace between the
//
and the first letter of the line comment. See the preceding example.Differentiate between
Property
and "property" in comments. They are different things.Property
is a type in AXON; property is any value associated with a JavaScript object. Often "field" can be used in exchange for "property" which can help with clarity.Files should be named like
CapitalizedCamelCasing.js
when returning a constructor, orlowerCaseCamelCasing.js
when returning a non-constructor function or singleton. When returning a constructor or singleton, the constructor name should match the filename.Every type, method and property should be documented.
The HTML5/CSS3/JavaScript source code must be reasonably well documented. This is difficult to specify precisely, but the idea is that someone who is moderately experienced with HTML5/CSS3/JavaScript can quickly understand the general function of the source code as well as the overall flow of the code by reading through the comments. For an example of the type of documentation that is required, please see the example-sim repository.
Assertions should be used appropriately and consistently. Type checking should not just be done in code comments. Use
Array.isArray
to type check an array.Abstract methods (normally implemented with an error) should be marked with
@abstract
jsdoc.If you need to namespace an inner class, use
{{namespace}}.register
, and include a comment about why the inner class needs to be namespaced. For example:Type Expressions
Type expressions should conform approximately to Google Closure Compiler syntax. PhET stretches the syntax in many cases (beyond the scope of this document to describe).
Prefer the most basic/restrictive type expression when defining APIs. For example, if a client only needs to know that a parameter is
{Node}
, don’t describe the parameter as{Rectangle}
.All method parameters should have type expressions. For example
@param {number} width
.In sim-specific code, options and fields should have type expressions when their type is not obvious from the context. “Obvious” typically means that the value type is clearly shown in the righthand-side of the definition. E.g.
const width = 42
clear shows thatwidth
is{number}
. E.g.const checkbox = new Checkbox(…)
clearly shows thatcheckbox
is{Checkbox}
. If the type is obvious from the context, the developer may still provide a type expression at his/her discretion. Examples:In common code repositories all options and fields should have type expressions, regardless of their visibility, and regardless whether their type is obvious from the context. If the same examples from above appeared in common code:
Type expressions for Enumeration values should be annotated as instances of that Enumeration, see examples in https://github.com/phetsims/phet-core/blob/master/js/Enumeration.js for more.
Look for cases where the use of type expressions involving Property subclasses are incorrect. Because of the structure of the
Property
class hierarchy, specifying type-specific Properties ({BooleanProperty}
,{NumberProperty}
,...) may be incorrect, because it precludes values of type{DerivedProperty}
and{DynamicProperty}
. Similarly, use of{DerivedProperty}
and{DynamicProperty}
precludes values of (e.g.){BooleanProperty}
. Especially in common code, using{Property,<TYPE>}
is typically correct, unless some specific feature of theProperty
subclass is required. For example,{Property.<boolean>}
instead of{BooleanProperty}
.Visibility Annotations
Because JavaScript lacks visibility modifiers (public, protected, private), PhET uses JSdoc visibility annotations to document the intent of the programmer, and define the public API. Visibility annotations are required for anything that JavaScript makes public. Information about these annotations can be found here. (Note that other documentation systems like the Google Closure Compiler use slightly different syntax in some cases. Where there are differences, JSDoc is authoritative. For example, use
Array.<Object>
orObject[]
instead ofArray<Object>
). PhET guidelines for visibility annotations are as follows:Use
@public
for anything that is intended to be part of the public API.Use
@protected
for anything that is intended for use by subtypes.Use
@private
for anything that is NOT intended to be part of the public or protected API.Put qualifiers in parenthesis after the annotation, for example:
@public (read-only)
. This indicates that the given property (AND its value) should not be changed by outside code (e.g. a Property should not have its value changed)@public (scenery-internal)
@public (a11y)
@public (phet-io)
@public (scenery-internal, read-only)
For JSDoc-style comments, the annotation should appear in context like this:
For Line comments, the annotation can appear like this:
Verify that every JavaScript property and function has a visibility annotation. Here are some helpful regular expressions to search for these declarations as PhET uses them.
x.y = something
:[\w]+\.[\w]+\s=
[\w]+: function\(
Math Libraries
DOT/Util.toFixed
orDOT/Util.toFixedNumber
should be used instead oftoFixed
. JavaScript'stoFixed
is notoriously buggy. Behavior differs depending on browser, because the spec doesn't specify whether to round or floor.IE11
Number.parseInt()
Array.prototype.find
Organization, Readability, and Maintainability
grunt find-duplicates
TODO
orFIXME
comments in the code? They should be addressed or promoted to GitHub issues.{{REPO}}Constants.js
file?PhetColorScheme
. Identify any colors that might be worth adding toPhetColorScheme
.DerivedProperty
instead ofProperty
?Accessibility
fuzzBoard&ea
)Node.accessibleOrder
used appropriately to maintain visual and PDOM layout balance?toUpperCase()
. Remember that one day these strings will be translatablePhET-iO
for the PhET-iO development process.
The text was updated successfully, but these errors were encountered: