fix installation of fish, enable autoconf configure argument

[norbusan]

No description provided.

[abraunegg]

LGTM though no way to test - I am assuming here that you have tested thus this is how this was found

[norbusan]

When building the Debian packages I realized that the fish completion was installed into $DESTDIR/@FISH_COMPLETION_DIR@ **literally**, so that must happen to each and everyone in fact ;-) -- --- PREINING Norbert                              https://www.preining.info Accelia Inc. + IFMGA ProGuide + TU Wien + JAIST + TeX Live + Debian Dev GPG: 0x860CDC13   fp: F7D8 A928 26E3 16A1 9FA0 ACF0 6CAC A448 860C DC13 Aug 11, 2020 19:00:03 abraunegg <notifications@github.com>:

…

@abraunegg approved this pull request. LGTM though no way to test - I am assuming here that you have tested thus this is how this was found — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub[#1016 (review)], or unsubscribe[https://github.com/notifications/unsubscribe-auth/AANHXJOUMBVPT3FGSRUOCOTSAEJCBANCNFSM4P2ROOJQ]. [https://github.com/notifications/beacon/AANHXJMVIL4AJDRY5HUAWN3SAEJCBA5CNFSM4P2ROOJ2YY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGODO3CJBI.gif]

[abraunegg]

When building the Debian packages I realized that the fish completion was installed into $DESTDIR/@FISH_COMPLETION_DIR@ literally, so that must happen to each and everyone in fact ;-) -- --- PREINING Norbert                              https://www.preining.info Accelia Inc. + IFMGA ProGuide + TU Wien + JAIST + TeX Live + Debian Dev GPG: 0x860CDC13   fp: F7D8 A928 26E3 16A1 9FA0 ACF0 6CAC A448 860C DC13 Aug 11, 2020 19:00:03 abraunegg notifications@github.com:
…
@abraunegg approved this pull request. LGTM though no way to test - I am assuming here that you have tested thus this is how this was found — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub[#1016 (review)], or unsubscribe[https://github.com/notifications/unsubscribe-auth/AANHXJOUMBVPT3FGSRUOCOTSAEJCBANCNFSM4P2ROOJQ]. [https://github.com/notifications/beacon/AANHXJMVIL4AJDRY5HUAWN3SAEJCBA5CNFSM4P2ROOJ2YY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGODO3CJBI.gif]

Ahh ok .. not good ...

Bump / re-release 2.4.4 with this fix as 2.4.5 ?

[SchoolGuy]

I also just noticed this while upgrading the Devel package. Sadly when approving PR #991 I only tested the completion and not the build process. Sorry about that.

Regarding the release... I think it would be good to have that fix in because it is broken in 2.4.4 but I leave that to you @abraunegg

[norbusan]

@SchoolGuy btw, what can be used as "default" installation directory? My patch uses /usr/loca/share/fish/completions but I'm not sure whether this actually would work.

[SchoolGuy]

According to https://fishshell.com/docs/current/#writing-your-own-completions this belongs into /usr/share/fish/completions which is the directory my completions are living in on openSUSE.

[norbusan]

Mumumu, but I really don't like it when packages with --prefix=/usr/local install files outside this hierarchy. This is very inpolite in my opinion (hating Gnome/Gtk for that). Furthermore, it probably depends on with what kind of prefix fish was compiled. If it was compiled with --prefix=/opt/fish I guess it will not search for /usr/share/fish/completions, or?

[norbusan]

Ahh, doing echo $fish_completion_path in a fish shell shows:

echo $fish_complete_path 
/home/norbert/.config/fish/completions /etc/fish/completions /home/norbert/.local/share/flatpak/exports/share/fish/vendor_completions.d /var/lib/flatpak/exports/share/fish/vendor_completions.d /usr/local/share/fish/vendor_completions.d /usr/share/fish/vendor_completions.d /usr/share/fish/completions /home/norbert/.local/share/fish/generated_completions

at least on Debian. That means that /usr/local/share/fish/vendor_completions.d might be better to use. WDYT?

[abraunegg]

@norbusan , @SchoolGuy

Do we need to do another PR here or is this just a packaging issue?

Do we release 2.4.5 to ensure there is no issue?

[norbusan]

I think we can leave everything as is. fish is underdocumented and it is not clear whether there is a global directory that is always searched.

We could require that -with-fish-completion-dir is given a path instead of allowing auto and yes (as we do for zsh and bash, where the directories can be found automatically or are well known).

But I don't think this is really an issue ... those who want fish need to pass a proper path to the configure call.

[abraunegg]

@norbusan

OK .. so .. do we do a 2.4.5 release to make sure everything is well rounded - as - if fish completions are added now, it installs to a poor location (that this PR fixes) ?

[norbusan]

Fine with me

[SchoolGuy]

@abraunegg Since we can override the directory via an option I am fine with 2.4.5

[SchoolGuy]

@norbusan Well my packaging experience is limited so I don't have such strong opinions about packages installing stuff somewhere on the system. But yes I agree that docs can never be extensive enough.

[abraunegg]

@norbusan , @SchoolGuy

Will do a 2.4.5 release then later tonight

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.