| Commit message (Collapse) | Author | Age |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes #5534.
Skylint was operating under assumption that all lvalues have a single
bound identifier, which is not true for `IndexExpression`s like
```
d["foo"] += "bar"
```
As a result, Skylint would crash. This change makes it handle cases without
bound identifiers gracefully.
Ideally `IndexExpression` assigments should be analyzed too, but it's a more
involved change.
Closes #5535.
PiperOrigin-RevId: 204109060
|
|
|
|
| |
PiperOrigin-RevId: 194985259
|
|
|
|
|
|
|
|
|
| |
This helps the Skylark interpreter to not depend on Bazel concepts, though it adds a temporary dependency of Skylint on packages/. The fix for that will be to create a Build API interface for BazelLibrary (e.g., "BazelLibraryAPI").
Refactored some GlobalFrame construction logic to be more uniform. Instead of constructing a whole Environment just to get a frame, we build the frame directly, using ImmutableMap.Builder to accumulate bindings. This convention may further change once we convert MethodLibrary and the like to @SkylarkGlobalLibrary, but for now it's more readable.
RELNOTES: None
PiperOrigin-RevId: 194960824
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 192472412
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 190071755
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 188225156
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 188217409
|
|
|
|
|
|
| |
RELNOTES:
Removed flag `--incompatible_load_argument_is_label`.
PiperOrigin-RevId: 187479614
|
|
|
|
|
|
|
| |
reporting that there is a missing 'Args:' section.
RELNOTES: None.
PiperOrigin-RevId: 186717757
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 186693205
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 186297559
|
|
|
|
|
|
|
|
|
|
|
|
| |
I neglected to unify the message text before submit. This uses the Go version
for both go and Java.
R=laurentlb,mkanat
CC=blaze-firehose+cl
RELNOTES:
unifly lint glob(["**/*.java"]) message
PiperOrigin-RevId: 185569422
|
|
|
|
|
|
|
|
|
|
|
| |
ControlFlowChecker has a precondition check that assumes nested functions do not occur. While this assumption is reasonable for a valid Skylark syntax, linter can actually be invoked on malformed files and users would get a stack trace instead of a human-readable linter error.
Alternative and possibly a better strategy would to not run CFChecker in case parse errors are detected.
fixes #4511
Closes #4512.
PiperOrigin-RevId: 185538897
|
|
|
|
|
|
|
|
|
| |
For now we will only block Java recursive globs. Any other languages or
extensions can be banned relatively easily.
RELNOTES:
Add lint check for discouraging glob(["**/*.java"])
PiperOrigin-RevId: 185417760
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 184667932
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In many cases, users do:
d = depset()
d += ...
To catch this issue, we use a heuristic to find which variable is a depset
(in theory, it could be reassigned - but it's unlikely and that would be
error-prone anyway)
RELNOTES: None.
PiperOrigin-RevId: 179536463
|
|
|
|
|
|
|
| |
Also, minor fix for the += operator check.
RELNOTES: None.
PiperOrigin-RevId: 177444203
|
|
|
|
|
|
|
| |
Split missing-docstring into missing-module-docstring and missing-function-docstring.
RELNOTES: None.
PiperOrigin-RevId: 176993716
|
|
|
|
|
| |
RELNOTES: None.
PiperOrigin-RevId: 176899549
|
|
|
|
|
|
|
|
| |
Blaze had its own class to avoid GC from varargs array creation for the precondition happy path. Guava now (mostly) implements these, making it unnecessary to maintain our own.
This change was almost entirely automated by search-and-replace. A few BUILD files needed fixing up since I removed an export of preconditions from lib:util, which was all done by add_deps. There was one incorrect usage of Preconditions that was caught by error prone (which checks Guava's version of Preconditions) that I had to change manually.
PiperOrigin-RevId: 175033526
|
|
|
|
|
|
|
| |
While at it, I added a flag "--single-file" that turns on single file mode. In this mode, only the specified file will be read (there is no dependency analysis).
RELNOTES: none
PiperOrigin-RevId: 174888506
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 174479316
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 174330237
|
|
|
|
|
|
|
| |
quotes
RELNOTES: None
PiperOrigin-RevId: 174319420
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 174318660
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 174310059
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 174305935
|
|
|
|
|
| |
RELNOTES: None
PiperOrigin-RevId: 174305030
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 173846739
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 173658526
|
|
|
|
|
|
|
| |
Mention the pattern for re-exporting symbols.
RELNOTES: none
PiperOrigin-RevId: 173658431
|
|
|
|
|
|
|
| |
While I was at it, I renamed a "*Tests" class to "*Test" for consistency.
RELNOTES: none
PiperOrigin-RevId: 173657530
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 173261316
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 173257316
|
|
|
|
|
|
|
| |
This used to be annoying for Skylint.
RELNOTES: none
PiperOrigin-RevId: 173249428
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 173140830
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 173125138
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 173113056
|
|
|
|
|
|
|
| |
In the process, I refactored and extended the line handling and error reporting a little.
RELNOTES: none
PiperOrigin-RevId: 172760285
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 172599144
|
|
|
|
|
|
|
| |
The end location now ends before the first statement of the function.
RELNOTES: none
PiperOrigin-RevId: 172578911
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It's good practice to place a "declaration" of a variable in a location that dominates all its uses, especially so if you want to document the variable. Hence we want no warning saying this declaration is unused.
Example:
var = None # don't warn about the unused declaration here
if condition:
var = 0
else:
var = 1
As noted in the code, this also has the undesired side-effect of suppressing the warning here:
var = None # doesn't warn either but ideally should
var = 0
Fixing the latter problem is not worth the increased code complexity, in my opinion.
RELNOTES: none
PiperOrigin-RevId: 172442402
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Global variables can have docstrings that are placed under their initialization.
Example:
variable = 0
"""Documentation for the variable.
Deprecated:
This variable is deprecated.
"""
Note that docstrings are not required, even for public variables. Hence there are no changes to the DocstringChecker.
RELNOTES: none
PiperOrigin-RevId: 172324350
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Each docstring should include
1. a summary line
2. a more extensive description (optional)
3. an 'Args' section (if there are parameters)
4. a 'Returns' section (if there is a return value)
5. a 'Deprecated' section (if it is deprecated)
in this order. If the order is different, there will be a warning.
The same order is recommended for Java (http://www.oracle.com/technetwork/java/javase/documentation/index-137868.html#orderoftags).
RELNOTES: none
PiperOrigin-RevId: 172313475
|
|
|
|
|
|
|
| |
Having a short canonical name for each kind of finding (the category) makes it easier to document all the lints and for users to find the documentation for each lint.
RELNOTES: none
PiperOrigin-RevId: 171972645
|
|
|
|
|
|
|
| |
Previously, the location of the warning was the whole file. Now it is 1:1-2:1. This doesn't make sense if the file is empty but I think it's good enough since that's an edge case.
RELNOTES: none
PiperOrigin-RevId: 171942703
|
|
|
|
|
|
|
| |
This only covers functions deprecated in the same file. It does not yet recognize load()ed functions that are deprecated in another file.
RELNOTES: none
PiperOrigin-RevId: 171841455
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 171532870
|
|
|
|
|
| |
RELNOTES: none
PiperOrigin-RevId: 171526636
|
|
|
|
|
|
|
|
| |
So far, only checks for usages of '+' on dictionary literals and
comprehensions.
RELNOTES: none
PiperOrigin-RevId: 170336917
|