Summary: Extend the `ArcanistXHPASTLinter::LINT_BRACE_FORMATTING` rule to raise a warning when `n_STATEMENT` is used without having `n_STATEMENT_LIST` as a parent. Essentially, this means that `if ($x) { do_y(); }` is preferred over `if ($x) do_y();`.
Test Plan: Added some test cases.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10571
Summary: Fixes T6627. This class documentation is out of date and somewhat misleading in the modern environment.
Test Plan: Read it.
Reviewers: btrahan, chad
Reviewed By: chad
Subscribers: epriestley
Maniphest Tasks: T6627
Differential Revision: https://secure.phabricator.com/D10899
Summary: This adjusts the behavior of PytestTestEngine so that it correctly reports test failures natively, rather than throwing a command error.
Test Plan: Yeah right
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10866
Summary: Extend `ArcanistXHPASTLinter::LINT_KEYWORD_CASING` to include the `class` keyword.
Test Plan: I can't really added a test case for this without getting "XHP19 Class-Filename Mismatch".
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D10653
Summary: Minor change, self-explanatory... this seems to make sense.
Test Plan: N/A
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10476
Summary: Adds an `ArcanistXHPASTLinter::LINT_CONSTRUCTOR_PARENTHESES` rule which ensures that (possibly empty) parentheses are used when calling a constructor.
Test Plan: Added test cases.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10597
Summary: In PHP 5.4, the `break` and `continue` statements no longer accept variable arguments (e.g., `break 1 + foo() * $bar;`). Static arguments still work, such as break 2;. As a side effect of this change `break 0;` and `continue 0;` are no longer allowed.
Test Plan: Added some test cases.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10567
Summary: It is not correct to add a trailing comma if the `n_ARRAY_VALUE` node is a heredoc string.
Test Plan: Added a test case.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10569
Summary: The `LINT_ARRAY_SEPARATOR` rule was added to `ArcanistXHPASTLinter` in D10535. Unfortunately, it does not correctly handle nested arrays correctly. This patch fixes this. Depends on D10554.
Test Plan: Added a test case.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10558
Summary:
Adds a rule to `ArcanistXHPASTLinter` which ensures that:
# Single-lined arrays //do not// contain an unnecessary trailing comma separator after the final array value.
# Multi-lined arrays //do// contain a trailing comma seperator after the final value value.
Depends on D10534.
Test Plan: Wrote and executed unit tests.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10535
Summary: I had assumed that we would always be interested in `n_STATEMENT_LIST`, which is fine for `if (...) { some_statements(); }` but not for `if (...) one_statement();`.
Test Plan: Added a test case.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10551
Summary: Improve the behavior of `ArcanistTextLinter` in dealing with `LINT_DOS_NEWLINE`. In particular, provide replacement text so that these linter issues can be autofixed.
Test Plan: Added a unit test.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10472
Summary: Explicitly list PHP magic methods (methods beginning with `__`) instead of assuming that //all// methods beginning with `__` are okay in terms of naming conventions. These magic methods were obtained from http://us1.php.net/manual/en/language.oop5.magic.php.
Test Plan:
```
Warning (XHP9) Naming Conventions
Follow naming conventions: methods should be named using lowerCamelCase.
1 <?php
2
3 class Foo {
>>> 4 function __foo() {}
5
6 function _bar() {}
7 }
Warning (XHP9) Naming Conventions
Follow naming conventions: methods should be named using lowerCamelCase.
3 class Foo {
4 function __foo() {}
5
>>> 6 function _bar() {}
7 }
```
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10539
Summary: Self-explanatory, improve test coverage.
Test Plan: Wrote and executed unit tests.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10474
Summary: It seems that `severity.rules` does not override the default linter severities set by the `ArcanistLinter` itself. This seems incorrect and counter-intuitive.
Test Plan: Created a `"severity.rules": { "(.*)": "disabled" }"` configuration and verified that it was being applied.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10475
Summary: D10132 improved this check by eliminating some false postives, but assumes an `n_FUNCTION_CALL` always starts with `n_SYMBOL_NAME`. This is true for `f()`, but not true for `C::m()` or `$v()`.
Test Plan: Added and ran unit tests.
Reviewers: btrahan, joshuaspence
Reviewed By: joshuaspence
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10467
Summary:
Fixes T5299. Currently, linting `src/future/http/HTTPSFuture.php` from rPHU will raise the following error:
```
Error (XHP31) Use Of PHP 5.3 Features
This codebase targets PHP 5.2.3, but `curlfile` was not introduced until
PHP 5.5.0.
532 // use this "@" stuff.
533
534 if (class_exists('CURLFile')) {
>>> 535 $file_value = new CURLFile((string)$tmp, $info['mime'], $info['name']);
536 } else {
537 $file_value = '@'.(string)$tmp;
538 }
```
However, since this class is being used conditionally, it //should// be fine and no linter errors should be being raised.
Test Plan: Added a test case.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Maniphest Tasks: T5299
Differential Revision: https://secure.phabricator.com/D10132
Summary: Fix linting for spacing around default parameter assignment in function/method declaration so that `function foo($x = null) {}` is preferred in favor of `function foo($x=null) {}`.
Test Plan: Modified existing unit tests.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10429
Summary: Adds an XHPAST linter rule for empty block statements. Basically, if a block statement is empty then it is much neater if the opening (`{`) and closing (`}`) braces are adjacent. Maybe this is just my own personal preference, in which case we could reduce the default severity to `ArcanistLintSeverity::SEVERITY_DISABLED`.
Test Plan: Wrote unit tests. I had to modify a bunch of existing unit tests accordingly.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10434
Summary: Add some missing function parameter typehints to the `ArcanistPhutilXHPASTLinter` class.
Test Plan: Eyeball it.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10430
Summary: `pht`ize some string in `ArcanistPhutilXHPASTLinter` so that they are translatable.
Test Plan: Not 100% sure that I am doing it right...
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10431
Summary: the call to setHeadCommit() was accidentally placed in a `if ($background) {}` block, which was removed in 54bea94. This adds the removed call.
Test Plan: used --head, worked as expected
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10447
Summary:
Calling `arc patch` on a diff that's dependent on a different diff tries to patch the parent first.
To patch the parent a child workflow is created, and a conduit is passed down, but the credentials are not and it is not marked as authenticated.
Then when the child tries to get the commit message for the dependency, it checks isConduitAuthenticated() https://secure.phabricator.com/diffusion/ARC/browse/master/src/workflow/ArcanistPatchWorkflow.php;2c3268f03ed70d3221eb1642bIc99ebb39b12902e$800 and on failure pops up an interactive editor for the commit message.
Instead we just pass down the credentials to the childred and mark them as authenticated, so this is not a problem.
Test Plan: With two diffs where DA2 depends on DA1, run `arc patch --force --nobranch DA2` ... this no longer pops an interactive editor for the commit message for the dependency.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: hach-que, seshness, epriestley, Korvin
Maniphest Tasks: T5986
Differential Revision: https://secure.phabricator.com/D10381
Summary: Update the `puppet-lint` linter to support the latest version (v1.0.1). This version allows a custom configuration file to be passed via `--config`, which brings this linter into line with other linters.
Test Plan: Update test cases to pass.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D10384
Summary: Adds a linter that checks php files for syntax errors using php -l
Test Plan:
Add a section to your .arclint file similar to:
```
"php": {
"type": "php",
"include": "(\\.php$)"
}
```
Then run arc lint on a php file with a syntax error.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, jacobwalker0814, Korvin
Differential Revision: https://secure.phabricator.com/D10370
Summary: flake8 v2.2.3 (released Aug 25) broke this. Just drop the stdin stuff.
Test Plan: User confirmed in IRC that this resolves the issue.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10358
Summary: Ref T5971. We lose validation of the line ranges, but I don't think that's a huge issue.
Test Plan: Ran `arc browse README`, `arc browse README:3`, and `arc browse README:3,6-7`
Reviewers: btrahan, epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Maniphest Tasks: T5971
Differential Revision: https://secure.phabricator.com/D10352
Summary: Ref T5781. This was broken in some earlier changes in T5781.
Test Plan: Ran `arc browse README`, `arc browse README:10`.
Reviewers: chad, btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T5781
Differential Revision: https://secure.phabricator.com/D10346
Summary: Fixes T3813. This error message isn't very helpful in SVN; be more helpful.
Test Plan: Ran `arc land` in a Subversion repository, got a reasonable error message.
Reviewers: btrahan, asherkin
Reviewed By: asherkin
Subscribers: aran
Maniphest Tasks: T3813
Differential Revision: https://secure.phabricator.com/D7275
Summary: Ref T5926. We only pass Phabricator an update hint from `arc land`, not from `arc commit`.
Test Plan:
- Ran `arc land` and `arc commit` with `--trace`, saw hints go over the wire.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T5926
Differential Revision: https://secure.phabricator.com/D10324
Summary:
Fixes T5870. When a filename contains spaces, Git will add a "\t" at the end of the "---" and "+++" lines. We currently consume this and think we're reading a file called "blah blah\t".
Instead, parse it out as not part of the filename.
Test Plan: Added failing unit test and made it pass.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T5870
Differential Revision: https://secure.phabricator.com/D10267
Summary:
This adds a lint engine for `hlint`, which is the standard and most general Haskell lint tool around these days.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
Test Plan: Install `hlint`, and run `arc unit`.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Projects: #arcanist
Differential Revision: https://secure.phabricator.com/D10250
Summary:
We currently require an exact number of "~" characters, but this is needless and error-prone.
Instead, allow any number larger than 3.
Test Plan:
- Added meta test-case.
- Ran `arc unit --everything`.
Reviewers: btrahan, thoughtpolice
Reviewed By: thoughtpolice
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10251
Summary:
Ref T4281. A long time ago, we added a `--background` flag to let `arc lint` and `arc unit` run while you're typing a commit message, in some situations.
This code is only moderately beneficial and is way too complicated. Particularly, it has a long history of causing hangs (T4281, T2463), doesn't work on Windows, and is impossible to debug.
It's also running into a serious PHP bug with EAGAIN/EPIPE being indistinguishable that I haven't been able to find a reasonable workaround for in ~3-4 hours of trying.
All the pathways forward that I can see make this already-complex system more complex.
The major reason that this stuff is so complex is that the subprocess may need to prompt the user (notably, to apply patches from lint).
Instead, I'm going to simplify how `arc diff` interacts with `arc lint` and `arc unit`, so we can just fire-and-forget a background process, let it do as much work as it can without needing user input, and then pick up wherever it left off. This will be slightly less cool/magical, but it won't hang bizarrely and I will be able to debug it.
For now, simply remove the `--background` flag and behavior so `arc` works for everyone.
Test Plan: Ran `arc diff` to create this diff.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T4281
Differential Revision: https://secure.phabricator.com/D10198
Summary: Ref T5781. `git show .` works like HEAD, but that isn't what `arc browse .` means.
Test Plan: Ran `arc browse .` with a repository at a published commit.
Reviewers: chad, btrahan, avive, avivey
Reviewed By: avivey
Subscribers: epriestley, avivey
Maniphest Tasks: T5781
Differential Revision: https://secure.phabricator.com/D10197
