Linguist: Incorrectly classifying files as OpenEdge ABL

[flipdazed]

The visual languages bar chart in the home page of this repository shows:

• OpenEdge ABL 72.4%
• Mathematica 17.6%
• Python 8.5%
• C 0.9%
• C++ 0.3%
• Matlab 0.3%

Clicking on the bar-chart and then on OpenEdge ABL sends the user to this page with

We couldn’t find any code matching ''

but displays what looks like a more correct list of languages as

Languages

• 139 Python
• 20 Markdown
• 15 Mathematica
• 12 C
• 6 CSV
• 4 Matlab
• 3 C++
• 1 Makefile
• 1 Text

[pchaigno]

The mismatch between the language bar and the search results looks like a caching issue (we've had a lot lately). It should update on the next push to the repository. If it does not, you can contact support so they trigger an update manually.

As for the 72.37% OpenEdge ABL, this happens because this file is detected as OpenEdge ABL. You can mark your data files as vendored using Linguist overrides to fix this.

[Ho-Holic]

I have similar behaviour in this repo. It treats TeX class files as OpenEdge ABL.

[Alhadis]

We probably need more samples for OpenEdge ABL. The ones we have now are pretty modest.

[pchaigno]

I have similar behaviour in this repo. It treats TeX class files as OpenEdge ABL.

@Ho-Holic Actually, the 47.08% TeX are due to this file.

[Ho-Holic]

But we are about wrong prediction of Open Edge ABL, aren't we?
This file is valid TeX class file, but it has OpenEdge ABL 43.2%

[pchaigno]

@Ho-Holic Oops, you're right. Sorry about that! Could you open a new issue?

We could probably add some heuristic rule to distinguish TeX from other languages with the .cls extension.

[Ho-Holic]

Done #3299