Arch Linux

raffaem

Member

Registered: 2023-07-04

Posts: 22

Requesting PKGBUILD public review

I request public review of my PKGBUILD: https://github.com/raffaem/pdf4qt-arch

My main concerns are:
1. What to do if upstream uses another package manager, like vcpkg, to manage dependencies. I skipped it and patched the make files to user Arch-provided libraries
2. What to do in case of missing CMake modules. I provide it in the package

Of course any other suggestions is more than welcome.

Trilby

Inspector Parrot

Registered: 2011-11-29

Posts: 30,361

Website

Re: Requesting PKGBUILD public review

My first notes:
- Get rid of all the empty variables.
- Why is there a GPL license in the PKGBUILD repo when the software itself uses LGPL3?
- It's missing a dependency on qt6-multimedia

To your questions:
1. Huh?  What package manager is used upstream is completely irrelevant - you write a package for arch, that's it.
2. This seems reasonable to me, but I have next to no expereince with cmake.

Last edited by Trilby (2023-07-04 21:24:46)

---

"UNIX is simple and coherent" - Dennis Ritchie; "GNU's Not Unix" - Richard Stallman

raffaem

Member

Registered: 2023-07-04

Posts: 22

Re: Requesting PKGBUILD public review

Thanks.

I fixed the first two.

It's missing a dependency on qt6-multimedia

Why do you say so? Namcapping the zst file shows no error.

Trilby

Inspector Parrot

Registered: 2011-11-29

Posts: 30,361

Website

Re: Requesting PKGBUILD public review

Perhaps it's only a build time dependency (which namcap cant detect) - but it was certainly required for me to build it.  Have you tried building it in a clean chroot?

Last edited by Trilby (2023-07-05 01:59:53)

---

"UNIX is simple and coherent" - Dennis Ritchie; "GNU's Not Unix" - Richard Stallman

a821

Member

Registered: 2012-10-31

Posts: 387

Re: Requesting PKGBUILD public review

Upstream has tags, contrary to this comment (copy-paste leftover?). You could use the third example from the wiki in `pkgver()`. (replace 'foo-' with 'v' to remove it from the tag names)

Edit: word

Last edited by a821 (2023-07-05 16:26:46)

raffaem

Member

Registered: 2023-07-04

Posts: 22

Re: Requesting PKGBUILD public review

Perhaps it's only a build time dependency (which namcap cant detect) - but it was certainly required for me to build it.  Have you tried building it in a clean chroot?

How do I do that?

I have read DeveloperWiki:Building_in_a_clean_chroot, but the "Convenience way" contains scripts for "building for a specific repository", where "repository" is one of {(core, extra, multilib) X (,-testing,-staging)}.

But I'm building for neither of those, I'm building for the AUR, right?

Should I use the "Classic way"?

Last edited by raffaem (2023-07-06 02:55:48)

Scimmia

Fellow

Registered: 2012-09-01

Posts: 12,650

Re: Requesting PKGBUILD public review

Since there are no built packages in the AUR, no, you're not building for it. You're building for yourself, it would be the same as if the package was going in extra, or extra-testing if you've got the testing repos enabled.

raffaem

Member

Registered: 2023-07-04

Posts: 22

Re: Requesting PKGBUILD public review

Since there are no built packages in the AUR, no, you're not building for it. You're building for yourself, it would be the same as if the package was going in extra, or extra-testing if you've got the testing repos enabled.

oh ok thanks

I run extra-x86_64-build, and found out I needed qt6-declarative and qt6-multimedia for the configuration process to succeed.

Then at the very end I got:

pdf4qt-git W: Dependency included, but may not be needed ('qt6-declarative')
pdf4qt-git W: Dependency included, but may not be needed ('qt6-multimedia')

Are those from namcap?
How can I reproduce them? Because namcapping the .pkg.tar.zst doesn't give me those messages.

Can I write in the wiki to use extra-x86_64-build when **making** a package for the AUR?

raffaem

Member

Registered: 2023-07-04

Posts: 22

Re: Requesting PKGBUILD public review

Sorry, since I found those out only by building in a clean chroot environment, and since namcapping the *.pkg.tar.zst file doesn't give me those as dependencies, these are makedepends, not depends, right?

raffaem

Member

Registered: 2023-07-04

Posts: 22

Re: Requesting PKGBUILD public review

pdf4qt-git is now in the AUR

raffaem

Member

Registered: 2023-07-04

Posts: 22

Re: Requesting PKGBUILD public review

Upstream has tags, contrary to this comment (copy-paste leftover?). You could use the third example from the wiki in `pkgver()`. (replace 'foo-' with 'v' to remove it from the tag names)

Edit: word

Thank you, I should have fixed it.

(1) Can you check again?
(2) Should I use "epoch=1" to force an update user-side, since I got the version numbering wrong the first time?

a821

Member

Registered: 2012-10-31

Posts: 387

Re: Requesting PKGBUILD public review

(1) You should have replaced 'VCS' with 'git' in `provides` and `conflicts` arrays, and in `prepare()`  `pkgver()` and `build()` you should use simple "$pkgname" (the substring '-VCS' does not exist in $pkgname so  "${pkgname%-VCS}" is unnecessary).

(2) Not needed, because "r850.ae3c43f" < "1.3.3.r10.gae3c43f" according to pacman rules (see vercmp(8)).

raffaem

Member

Registered: 2023-07-04

Posts: 22

Re: Requesting PKGBUILD public review

(1) You should have replaced 'VCS' with 'git' in `provides` and `conflicts` arrays, and in `prepare()`  `pkgver()` and `build()` you should use simple "$pkgname" (the substring '-VCS' does not exist in $pkgname so  "${pkgname%-VCS}" is unnecessary).

(2) Not needed, because "r850.ae3c43f" < "1.3.3.r10.gae3c43f" according to pacman rules (see vercmp(8)).

Done, thanks!