You are not logged in.

#1 2022-08-28 18:39:35

xerus
Member
Registered: 2021-05-11
Posts: 20

updpkgsums does not honor SKIP entries

Whenever I run updpkgsums, it replaces SKIP with a checksum, which is not why I entered SKIP for that particular artifact.
This is kinda annoying, is there a way around that?

Last edited by xerus (2022-08-28 18:39:52)

Offline

#2 2022-08-28 18:42:53

jasonwryan
Anarchist
From: .nz
Registered: 2009-05-09
Posts: 30,191
Website

Re: updpkgsums does not honor SKIP entries

Paste the PKGBUILD where this happens.


Arch + dwm   •   Mercurial repos  •   Surfraw

Registered Linux User #482438

Offline

#3 2022-08-31 12:38:04

xerus
Member
Registered: 2021-05-11
Posts: 20

Re: updpkgsums does not honor SKIP entries

Offline

#4 2022-08-31 12:42:18

Alad
Wiki Admin/IRC Op
From: Bagelstan
Registered: 2014-05-04
Posts: 2,359
Website

Re: updpkgsums does not honor SKIP entries

It has a pkgver() function, the output will be used to determine the value of pkgver. Not sure why you're trying to set SKIP in the first place - this would remove any kind of integrity checks.


Mods are just community members who have the occasionally necessary option to move threads around and edit posts. -- Trilby

Offline

#5 2022-08-31 17:15:11

a821
Member
Registered: 2012-10-31
Posts: 347

Re: updpkgsums does not honor SKIP entries

If you really want to SKIP checksums, then, why do you need to run updpkgsums in the first place?


I cannot find in the pacman docs where it says that if you put 'SKIP' in the checksum array, the generation of the chksum will be skipped, only that the verification will be skipped, not the same thing.

PKGBUILD(5) wrote:

       cksums (array)
           This array contains CRC checksums for every source file specified in the source array (in the same order). makepkg will use this to verify source file integrity
           during subsequent builds.If SKIP is put in the array in place of a normal hash, the integrity check for that source file will be skipped. To easily generate
           cksums, run “makepkg -g >> PKGBUILD”. [...]

updpkgsums calls "makepkg -g" internally, and the function that generates the checksums is this one (I suppose): generate_one_checksum

If I understand it correctly, and correct me if I'm wrong, the generation is skipped only for signature files and VCS sources. Nowhere it checks for SKIP.

Offline

#6 2022-08-31 17:29:35

Scimmia
Fellow
Registered: 2012-09-01
Posts: 9,555

Re: updpkgsums does not honor SKIP entries

That's not even a very functional PKGBUILD.

You need to version the source and add a checksum instead of skipping it. That means that the original question is moot.

You also can't install things to /usr/local/. Exporting a variable in the .install file is useless and wrong. You shouldn't be using replaces.

Last edited by Scimmia (2022-08-31 17:30:08)

Offline

#7 2022-09-10 12:16:49

xerus
Member
Registered: 2021-05-11
Posts: 20

Re: updpkgsums does not honor SKIP entries

1. My tooling runs updpkgsums automatically. And what if there are multiple sources, and I need to skip the checksum on one?
2. I can't version the source, it is not under my control and updated every few days.
3. I used `replaces` because I renamed my package.
4. Why can't I install to `/usr/local`? It works, and was the only way I found so far to override texlive binaries without conflict.
5. I am removing the .install file, it was dated anyways.

Offline

#8 2022-09-10 12:41:47

Trilby
Inspector Parrot
Registered: 2011-11-29
Posts: 27,272
Website

Re: updpkgsums does not honor SKIP entries

xerus wrote:

4. Why can't I install to `/usr/local`? It works, and was the only way I found so far to override texlive binaries without conflict.

See the very first bullet.  While I don't know if it'd be the right approach for this package, installing under /opt/ is perfectly acceptable for packages.

EDIT: the license seems wrong.  I was first going to suggest this package should have a -bin suffix as it installs precompiled binaries, but I took this point back as that rule only applies to precompiled binaries for which the source code is available.  If the source code for this package is available, you may just want to use that, if it isn't, it can't be GPL licensed.  So at least one of these must change.

EDIT 2: Despite not distributing source code or making it available anywhere, upstream does claim to license this as GPL ... though they repeatedly call it the "Creative Commons GNU GPL", and then admit:

upstream genius wrote:

In practice, users may forget about the legal part, if only because I haven't even read (and understood) it completely myself

Well that much is obvious.  Please do not post this in the AUR until upstream pulls their head out of their ass figures out what license they really want.  But then again, they seem to like their head where it is:

... and don't bother discussing licence issues and related things with us for the mere sake of discussing licence stuff.

EDIT 3: another concern is that the source array doesn't actually include the source material but apparently is just the "installer" which when run needs network access to download all the actual content.  This is not appropriate for PKGBUILDs.  There should be no requirement for a network connection in the build and package functions.

Last edited by Trilby (2022-09-10 13:41:25)


"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" -  Richard Stallman

Online

Board footer

Powered by FluxBB