linter: make constraint-name-style configurable #2017

IEncinas10 · 2023-09-23T20:33:53Z

Up until now, constraint-name-style rule required constraint names to end in _c. This makes it configurable, so that users can require constraint names to start with c_ instead.

I'm open to sugestions about the configuration parameter's name, it doesn't convince me too much but I can't think of a better option. We could have check_suffix and check_prefix but as it doesn't make sense for them to be on at the same time I don't like it too much either.

Edit: forgot to handle this, leaving it here so I don't forget

Handle invalid regex provided by user

codecov-commenter · 2023-09-23T20:51:24Z

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (a4d61b1) 92.95% compared to head (846710b) 92.95%.

Files	Patch %	Lines
...og/analysis/checkers/constraint_name_style_rule.cc	94.11%	1 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2017      +/-   ##
==========================================
- Coverage   92.95%   92.95%   -0.01%     
==========================================
  Files         358      359       +1     
  Lines       26446    26458      +12     
==========================================
+ Hits        24583    24594      +11     
- Misses       1863     1864       +1

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

hzeller · 2023-12-19T15:27:35Z

(Sorry for the long delay, I was pretty busy with other things...)

Maybe we can provide a regular expression, e.g. configuration parameter name pattern, and then the user can decide. And then we can set the default to "([a-z0-9]+_)+c" or so to enforce snake-case with trailing-_c' (we can keep the regex simple and don't have to worry about the beginning starting with a digit as that already has been taken care of by the lexer/parser)

Using regular expressions is probably also a good choice to configure some of the other rules - they are
We use the re2 library successfully in other parts of the project e.g. in ./common/tools/jcxxgen.cc and ./verilog/tools/ls/autoexpand.cc.

IEncinas10 · 2023-12-20T19:54:55Z

No worries, first things first!

I remember seeing discussion about regular expressions for some other rules in the past but thought the support wasn't there yet. I'll give this a look and definitely change it. If it ends up being as easy at it sounds I'll make an issue to eventually update every rule that could make use of this feature.

hzeller · 2023-12-21T11:24:27Z

At the time we had the disccussions about regular expressions, the choice was std::regex, but that is generally a poor choice as it is slow. Since then, RE2 had been added to the dependencies, and that is pretty performant.

IEncinas10 · 2024-01-13T16:57:26Z

verilog/analysis/checkers/constraint_name_style_rule.cc

-static const Matcher& ConstraintMatcher() {
+absl::Status ConstraintNameStyleRule::Configure(
+    absl::string_view configuration) {
+  std::string pattern = kSuffix;


I think this is redundant but the rationale is the following:

If at some point we add other parameters to the rule, but pattern is not provided, the ParseNameValues wouldn't set any value to pattern, so we would then need to set a default value.

Avoid future problems by doing it already?

IEncinas10 · 2024-01-13T17:02:08Z

verilog/analysis/checkers/constraint_name_style_rule.cc

-      violations_.insert(LintViolation(*identifier_token, kMessage, context));
+    if (!RE2::FullMatch(constraint_name, *regex)) {
+      violations_.insert(
+          LintViolation(*identifier_token, FormatReason(), context));


Didn't find many resources on RE2. Can it somehow help us provide a better linting message than just "follow the regex"? Perhaps provide an autofix?

hzeller · 2024-11-21T22:03:25Z

Looking backwards in time to see if there are active PRs. This looks like it is still relevant though probably some rebase needed due to conflicts.

IEncinas10 · 2024-11-23T10:47:53Z

Looking backwards in time to see if there are active PRs. This looks like it is still relevant though probably some rebase needed due to conflicts.

Updated, thanks for the reminder!

hzeller · 2024-11-26T17:50:33Z

verible/verilog/analysis/checkers/constraint-name-style-rule.h

  void HandleSymbol(const verible::Symbol &symbol,
                    const verible::SyntaxTreeContext &context) final;

  verible::LintRuleStatus Report() const final;

+  std::string Pattern() const { return regex->pattern(); }


Is this public method (returning a string by value) needed anywhere ?

It isn't, I used it to diagnose if the user-provided regex failed to parse but it is not needed anymore thanks to the SetRegex utility. Missed it, thanks!

hzeller · 2024-11-26T17:51:22Z

verible/verilog/analysis/checkers/constraint-name-style-rule.h

-// convention.
-// The constraints should be named with lower_snake_case and end with _c.
+// Lower snake case, ends with `_c`
+#define kSuffix "([a-z0-9]+_)+c"


Instead of a macro, can this be a static constexpr absl::string_view kSuffix in the private section of ConstraintNameStyleRule instead ?

Indeed, thanks! Before kPrefix was here too and that "needed" to be a define to do this stupid trick https://github.com/IEncinas10/verible/blob/de368dbc5d0a46ad14763854ee105f89a65d476a/verible/verilog/analysis/checkers/constraint-name-style-rule_test.cc#L73

verible/verilog/analysis/checkers/constraint-name-style-rule.h

hzeller · 2024-11-26T19:17:34Z

verible/verilog/analysis/checkers/constraint-name-style-rule_test.cc

 using verible::RunLintTestCases;

+// Lower snake case, starts with `c_`
+#define kPrefix "c+(_[a-z0-9]+)+"


I'd probably move this into the VariousPrefixTests TEST to be close to where it is needed.
(and since it is only used in one place to assemble the pattern: configuration, maybe it doesn't need to
be a macro at all).

(other than that, looks great and ready to be merged)

I agree. I needed this because my original intent was to offer a choice between _c suffix and prefix. Thanks!

Make constraint-name-style rule regex-configurable while maintaining its default behavior of requiring constraint names to be lower snake case ending in `_c`

hzeller

Nice.
Thanks!

IEncinas10 mentioned this pull request Jan 13, 2024

Migrate name style rules to regex #2074

Open

10 tasks

IEncinas10 force-pushed the configurable-constraint-name-style-rule branch from 5e9b7da to bbccb82 Compare January 13, 2024 16:52

IEncinas10 commented Jan 13, 2024

View reviewed changes

IEncinas10 force-pushed the configurable-constraint-name-style-rule branch 2 times, most recently from 7332d1b to 846710b Compare January 14, 2024 10:24

IEncinas10 requested a review from hzeller March 29, 2024 13:25

hzeller added the want-merge-needs-rebase a pull request that should not be lost, but needs rebasing first. label Nov 23, 2024

IEncinas10 force-pushed the configurable-constraint-name-style-rule branch from 846710b to de0dffd Compare November 23, 2024 10:45

IEncinas10 force-pushed the configurable-constraint-name-style-rule branch 3 times, most recently from f97ec65 to d666d5e Compare November 26, 2024 17:46

hzeller reviewed Nov 26, 2024

View reviewed changes

verible/verilog/analysis/checkers/constraint-name-style-rule.h Show resolved Hide resolved

IEncinas10 force-pushed the configurable-constraint-name-style-rule branch from d666d5e to de368db Compare November 26, 2024 18:47

hzeller reviewed Nov 26, 2024

View reviewed changes

linter: make constraint-name-style configurable

2191f11

Make constraint-name-style rule regex-configurable while maintaining its default behavior of requiring constraint names to be lower snake case ending in `_c`

IEncinas10 force-pushed the configurable-constraint-name-style-rule branch from de368db to 2191f11 Compare November 27, 2024 16:36

hzeller approved these changes Nov 27, 2024

View reviewed changes

hzeller merged commit f3da2ce into chipsalliance:master Nov 27, 2024
34 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

linter: make constraint-name-style configurable #2017

linter: make constraint-name-style configurable #2017

IEncinas10 commented Sep 23, 2023 •

edited

Loading

codecov-commenter commented Sep 23, 2023 •

edited

Loading

hzeller commented Dec 19, 2023

IEncinas10 commented Dec 20, 2023

hzeller commented Dec 21, 2023

IEncinas10 Jan 13, 2024

IEncinas10 Jan 13, 2024

hzeller commented Nov 21, 2024

IEncinas10 commented Nov 23, 2024

hzeller Nov 26, 2024

IEncinas10 Nov 26, 2024

hzeller Nov 26, 2024

IEncinas10 Nov 26, 2024

hzeller Nov 26, 2024

hzeller Nov 26, 2024

IEncinas10 Nov 27, 2024

hzeller left a comment

linter: make constraint-name-style configurable #2017

linter: make constraint-name-style configurable #2017

Conversation

IEncinas10 commented Sep 23, 2023 • edited Loading

codecov-commenter commented Sep 23, 2023 • edited Loading

Codecov Report

hzeller commented Dec 19, 2023

IEncinas10 commented Dec 20, 2023

hzeller commented Dec 21, 2023

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hzeller commented Nov 21, 2024

IEncinas10 commented Nov 23, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hzeller left a comment

Choose a reason for hiding this comment

IEncinas10 commented Sep 23, 2023 •

edited

Loading

codecov-commenter commented Sep 23, 2023 •

edited

Loading