Add .cnf to INI

[eggplants]

Description

I would like to add .cnf extension as INI.

.cnf is used for MySQL Option File and OpenSSL CONF library configuration file.

Checklist:

• I am associating a language with a new file extension.
  • The new extension is used in hundreds of repositories on GitHub.com
    • Search results for each extension:
      • https://github.com/search?q=path%3A*.cnf+mysqld+NOT+nothack&type=code
    • I have included a real-world usage sample for all extensions added in this PR:
    • Sample source(s):
      • https://github.com/mysql/mysql-shell/blob/037114f/unittest/data/config/my.cnf
        • GNU General Public License, version 2.0
      • https://github.com/nodejs/node/blob/66da32c/deps/openssl/openssl/apps/openssl.cnf
        • MIT License
  • I have included a change to the heuristics to distinguish my language from others using the same extension.

[Alhadis]

Distribution appears to check out (searching for .cnf files yields ~109k results, most of which appear to be MySQL configs.

[lildude]

Can you please add another .cnf sample, maybe one demonstrating the OpenSSL conf file? This one sample you've added seems to be influencing the classifier such that it is now detecting a .pro file as INI:

Proguard/proguard-rules2.pro BAD (INI)

Hopefully another sample will be enough to correct this 🤞

[lildude]

😢 that didn't do the trick so I've bumped the number of allowed failures.

[Alhadis]

@lildude Please ⌘+R my memory: does Linguist's classifier take comments into consideration when analysing keyword frequency? Because the long-winded legalese preceding every GPL-licensed bit of code might skew the classifier's judgement if another candidate language's samples are also half-lawyer-speak, half actual source-code.

The problematic sample you mentioned earlier began with the GPL, so it left me wondering…

[lildude]

Please ⌘+R my memory: does Linguist's classifier take comments into consideration when analysing keyword frequency?

Yes, and thanks to the tokenizer, it detects it as a comment so ignores things after the hash when classifying. With LINGUIST_DEBUG=2 you'll see lines like this (I forced this by adding .cnf to the Proguard entry):

$ LINGUIST_DEBUG=2 bundle exec bin/github-linguist samples/INI/my.cnf
samples/INI/my.cnf: 107 lines (93 sloc)
  type:      Text
  mime type: text/plain
                    #       INI  Proguard
               #   19    29.260         -
              ##    1         -     4.395
               -   18         -    28.547
              .7    1     6.651         -
            .log    1     6.651         -
            .pid    1     6.651         -
           .sock    3    23.248         -
               /   28   302.228         -
               :    2         -     1.737
               ;    2         -     2.936
               =   47   612.767         -
          AES256    1     6.651         -
               C    1     7.344         -
     CAMELLIA256    1     6.651         -
        COMMENT#   28     1.506         -
        COMMENT;    1     8.848         -
CaseSensitiveOpt    1     6.651         -
[...]
win_path_with_po    1     6.651         -
       INI =  -2044.596 +  -5.312 =  -2049.908
  Proguard =  -4755.497 +  -6.852 =  -4762.349
  language:  INI
$

Stripping the GPL license comment makes no difference:

$ LINGUIST_DEBUG=2 bundle exec bin/github-linguist samples/INI/my.cnf
samples/INI/my.cnf: 107 lines (93 sloc)
  type:      Text
  mime type: text/plain
                    #       INI  Proguard
               #   19    29.260         -
              ##    1         -     4.395
               -   18         -    28.547
              .7    1     6.651         -
            .log    1     6.651         -
            .pid    1     6.651         -
           .sock    3    23.248         -
               /   28   302.228         -
               :    2         -     1.737
               ;    2         -     2.936
               =   47   612.767         -
          AES256    1     6.651         -
               C    1     7.344         -
     CAMELLIA256    1     6.651         -
        COMMENT#   28     1.506         -
        COMMENT;    1     8.848         -
CaseSensitiveOpt    1     6.651         -
[...]
win_path_with_po    1     6.651         -
       INI =  -2044.596 +  -5.312 =  -2049.908
  Proguard =  -4755.497 +  -6.852 =  -4762.349
  language:  INI
$

The problematic sample you mentioned earlier began with the GPL, so it left me wondering…

Yeah, this shouldn't make a difference as pointed out above, but I removed the GPL license text and tested again locally and it didn't make a difference. I still got the same failure that needed the number bump.