[aur-general] AUR4 keywords in PKGBUILD?
Is it possible currently to set the keywords/tags in the PKGBUILD itself instead of on the website? (Previously, we had to set categories on the AUR website, but that felt suboptimal.) e.g. a keywords=('cats' 'dogs' ...); variable? Or, alternatively, would people be amenable to adding an ssh command in the AUR git-shell to add/remove/set keywords?
On Sun, 12 Jul 2015 at 18:07:37, Ido Rosen wrote:
Is it possible currently to set the keywords/tags in the PKGBUILD itself instead of on the website? (Previously, we had to set categories on the AUR website, but that felt suboptimal.) e.g. a keywords=('cats' 'dogs' ...); variable?
No.
Or, alternatively, would people be amenable to adding an ssh command in the AUR git-shell to add/remove/set keywords?
Sounds like a nice suggestion to me. Could you please file a FR on the bug tracker?
On 12/07, Lukas Fleischer wrote:
On Sun, 12 Jul 2015 at 18:07:37, Ido Rosen wrote:
Is it possible currently to set the keywords/tags in the PKGBUILD itself instead of on the website? (Previously, we had to set categories on the AUR website, but that felt suboptimal.) e.g. a keywords=('cats' 'dogs' ...); variable?
No.
I wonder if it would be useful to support having keywords in the .SRCINFO file. A couple of people have requested it, though I'm still unsure about the idea. What do you think?
Or, alternatively, would people be amenable to adding an ssh command in the AUR git-shell to add/remove/set keywords?
Sounds like a nice suggestion to me. Could you please file a FR on the bug tracker?
-- Sincerely, Johannes Löthberg PGP Key ID: 0x50FB9B273A9D0BB5 https://theos.kyriasis.com/~kyrias/
On Mon, Jul 13, 2015 at 3:45 PM, Johannes Löthberg <johannes@kyriasis.com> wrote:
On 12/07, Lukas Fleischer wrote:
On Sun, 12 Jul 2015 at 18:07:37, Ido Rosen wrote:
Is it possible currently to set the keywords/tags in the PKGBUILD itself instead of on the website? (Previously, we had to set categories on the AUR website, but that felt suboptimal.) e.g. a keywords=('cats' 'dogs' ...); variable?
No.
I wonder if it would be useful to support having keywords in the .SRCINFO file. A couple of people have requested it, though I'm still unsure about the idea.
What do you think?
Definitely worthwhile.
Or, alternatively, would people be amenable to adding an ssh command in the AUR git-shell to add/remove/set keywords?
Sounds like a nice suggestion to me. Could you please file a FR on the bug tracker?
-- Sincerely, Johannes Löthberg PGP Key ID: 0x50FB9B273A9D0BB5 https://theos.kyriasis.com/~kyrias/
I wouldn't say having keywords in the .SRCINFO is any better than having it in the PKGBUILD. SRCINFOs contain metadata drawn directly from the PKGBUILD, with the exception of things like the `epoch` field. To have the keywords present in the SRCINFO but not the PKGBUILD would be nonsensical in my opinion.
On 14/07, David Phillips wrote:
I wouldn't say having keywords in the .SRCINFO is any better than having it in the PKGBUILD. SRCINFOs contain metadata drawn directly from the PKGBUILD, with the exception of things like the `epoch` field. To have the keywords present in the SRCINFO but not the PKGBUILD would be nonsensical in my opinion.
The AUR does not parse, create, nor care about PKGBUILDs other than that they exist in the package. -- Sincerely, Johannes Löthberg PGP Key ID: 0x50FB9B273A9D0BB5 https://theos.kyriasis.com/~kyrias/
I'm aware of that and I hope I didn't imply otherwise. The point I was trying to get at is that .SRCINFOs are summaries of a package's metadata. I understand that keywords are metadata so if they had to be put somewhere, it would make sense to put them in .SRCINFOs. However (assuming we refuse to allow them into a PKGBUILD) this would require another input to the SRCINFO other than the PKGBUILD. Assuming this input took the form of another separate file, it would be unnecessary clutter. I suppose if mksrcinfo left the proposed keyword field alone but updated the rest of the file, it might be a tidier solution…
* David Phillips <dbphillipsnz@gmail.com> (Sat, 18 Jul 2015 10:25:43 +1200):
I suppose if mksrcinfo left the proposed keyword field alone but updated the rest of the file, it might be a tidier solution…
I'm confused: do you want to say that you want a human to fill in keywords in .SRCINFO and let mksrcinfo fill in the blanks? I fear that is error prone, as a user might not type the info in a way the parser expects it. Best, Marcel
participants (5)
-
David Phillips
-
Ido Rosen
-
Johannes Löthberg
-
Lukas Fleischer
-
Marcel Korpel