gpg.fish refactor

[LawAbidingCactus]

Description

Refactors gpg.fish to conditionally activate completions based on gpg version, as discussed in #6062:

The following was discussed briefly in #2401, but remained unresolved. On many systems, gpg is symlinked to gpg2-- when this is the case, it would be more appropriate to use these completions than fish's current ones (with gpg2 in the completion script replaced with gpg). gpg.fish could test for the GPG version and activate version-specific completions depending on the result.

TODOs:

• Changes to fish usage are reflected in user documentation/manpages.
• Tests have been added for regressions fixed
• User-visible changes noted in CHANGELOG.md

[zanchey]

As a general overview, I don't think this is the right approach - there is lots of duplicated code which I think could be simplified.

I wonder about:

• breaking gpg2 completions out into __fish_complete_gpg2, which takes a __fish_complete_gpg2_command argument, and which uses $__fish_complete_gpg2_command instead of gpg2 everywhere (in functions and completion arguments
• in gpg2.fish, call __fish_complete_gpg_command gpg2
• in gpg.fish, check if gpg is 2.x, and if so call __fish_complete_gpg2 gpg; otherwise use the gpg1 completions

Would that work?

[ridiculousfish]

I merged this but then saw the comments, so reverted the merge.

[LawAbidingCactus]

@ridiculousfish: Yeah, sorry-- I implemented @zanchey's idea for deduplicating redundant completion code and decided to see if I could take it a little further. It's not ready yet, since I haven't had time to work on it lately.

[LawAbidingCactus]

I've rewritten the PR to deduplicate redundant code to the fullest extent possible. The contextual completion functions, along with the GPG completions themselves, have been moved to `share/fish/functions/`. This allows `__fish_complete_gpg` to be called from the actual completion files to conditionally generate completions from the GnuPG version and argument given. This also allows completions for `gpg1` (as in, a binary called `gpg1`) to be easily generated, without adding redundant completion code. Some commenting is required but I'm otherwise done.

…

On Sun, Oct 6, 2019 at 19:36, ridiculousfish ***@***.***> wrote: I merged this but then saw the comments, so reverted the merge.505793 — You are receiving this because you authored the thread. Reply to this email directly, [view it on GitHub](#6139?email_source=notifications&email_token=AB7UAJG2MOEU5BTEOI4ORR3QNI423A5CNFSM4I2FWOMKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEAOSEHY#issuecomment-538780191), or [mute the thread](https://github.com/notifications/unsubscribe-auth/AB7UAJCX5RSPUC6QASKLKK3QNI423ANCNFSM4I2FWOMA).

[ridiculousfish]

note zanchey is on vacation so this may sit a while. If you are keen to have this merged before their return let me know.

[zanchey]

Looks great; one comment-related change but otherwise good to go.

[krobelus]

Merged as 21a6a19, thanks for all the work!

[zanchey]

Thanks for merging @krobelus, I was hoping to get to this today.