Linguist: Perl 5 code being mistaken for Perl 6

[wchristian]

This is recognized as Perl 6, wrongly:

=head1 NAME

class

=cut

package PH;

use 5.010;

This is recognized as Perl 5:

=head1 NAME



=cut

package PH;

use 5.010;

[wchristian]

I don't get reliable results when i push this to github with changes, but this looks like it also is recognized as Perl 6, wrongly:

use 5.006;
use strict;
use warnings;

=head1

module

=cut

And removing module changes it to perl 5.

[wchristian]

Forgot to mention, in both examples the files were Word.pm.

[zoffixznet]

Looks like just swapping these conditionals to check for Perl 5 regex first will solve the issue.

And then the cases where the disambiguation gets it wrong, Perl 6 users always have the option to renaming the file to .pm6 extension.

[kentfredric]

https://github.com/github/linguist/blob/58ab593a64088be56e3323a9fb121fe0988bdb3c/lib/linguist/heuristics.rb#L348

^ That heuristic would false positive on anything using MooseX::Declare style perl-5 syntax anyway.

So yeah, +1 on inversion.

[pchaigno]

We'd welcome a pull request with a test (test/test_heuristics.rb) to implement that change :smiley:

[smola]

Is it really worth it to have Perl 6 separated from Perl? It seems the only case where linguist tries to determine language version, no Python 3, Java 6 or things like that, and the amount of misclassifications seems significant.

[zoffixznet]

Is it really worth it to have Perl 6 separated from Perl

Yes, considering it's an entirely different language. It's like C vs C++, except with poorer name.

[Alhadis]

@smola Perl 5 and Perl 6 are, as @zoffixznet put it, two entirely separate languages. To quote "Programming Perl":

That 6 is deceptive though; Perl 6 is really a “kid sister” language to Perl 5, and not just a major update to Perl 5 that version numbers have trained you to expect.

[tessarin]

@pchaigno Do you know if this was in fact fixed? I've got a similar problem in a repository for a Perl 5 module which has its main file being incorrectly interpreted as Perl 6, making the repository also Perl 6 (funny enough, named p5-POD-Auto). Thanks!

[pchaigno]

@tessarin No, I don't think a pull request for the above proposed change was submitted.

[tessarin]

@pchaigno May I submit one then? You mentioned adding a test to test/test_heuristics.rb, however there is already a test for loading samples and disambiguate _Perl_ | _Perl 6_ (lines 224-231). In this case, adding a minimal example to samples/Perl (like the ones posted here) causes tests to fail. With the change in heuristics.rb they pass again. So, is adding the sample enough?

[pchaigno]

May I submit one then?

@tessarin Sure, go ahead 😄

You mentioned adding a test to test/test_heuristics.rb, however there is already a test for loading samples and disambiguate _Perl_ | _Perl 6_ (lines 224-231).

Oh, right. In that case, yes, a sample is enough. Please add it in test/fixtures/Perl/ rather than in samples/ though. samples/ files should originate from "real" application. test/fixtures/ is for this kind of test-only example files.

[sidyll]

Any chance that this will get fixed soon?

[tessarin]

@sidyll well the pull request was already made and it was approved, but I'm not sure about how the review process works or the release frequency.

[lildude]

PR has been merged. I've been AFK for a week so need to do some catching up before I can make a new release. I can't say when this will be right now though.

[tessarin]

Thank you @lildude!