GitHub thinks Octo files are Roff files

[tobiasvl]

According to GitHub, this repo consists almost entirely of Roff files.

We could consider fixing this by adding Octo support to Linguist, the library that detects languages for GitHub. So if that issue is closed, this one can be closed too.

But if we can't be bothered to do that, this repo should probably override the filetype manually.

[benzammour]

As the referenced and this issues are almost 3 years old, I assume adding support to Linguist was not successful?
If so, is it okay if I add a PR that does this manually?

[JohnEarnest]

@benzammour OK, I'm fine with that.

It's possible that we stand a better chance getting recognized as a valid file extension today than we did 3 years ago, though. Many more repositories containing Octo programs or tools compatible with Octo assembly language exist now.

[benzammour]

After doing some research, this is not as trivial as it seems.

Firstly, Octo does not meet the "at least 2000 files per extension indexed in the last year" requirement defined here.
Additionally, as we are not able to override the settings to our own language but are restricted to pre-defined ones, we would only be able to use something like Assembly.

However, this is also not desirable, therefore, this issue cannot be resolved trivially

[JohnEarnest]

A different avenue which might at least edge us mildly closer to what we want would be to argue for a curtailment of roff extensions. Last time I investigated, it took claim of an absolutely absurd set of wildcard patterns which I doubt reflect actual usage.

[JohnEarnest]

Just to add some concrete statistics to the discussion, and for posterity, searching GitHub by extension for .8o:

https://github.com/search?q=extension%3A8o&type=code

Presently produces 822 results. I took a few wild stabs before running into API rate limits and I didn't find a single result that wasn't Octo assembly.