Arch Linux User Repository

@Rhinoceros - i was going over the code yesterday and ran into 2 issues. the first is what to do about the context parameter required by the callback functions (e.g. dlcb = download callback), mentioned already by @Thulinma. i can test it with a null context, if i could compile everything else. which brings me to the 2nd issue - kalu-dbus.c references a number of EVENT enumerations, not all of which match the old ones. these 3 i can match up:

ALPM_EVENT_PKGDOWNLOAD_START -> ALPM_EVENT_PKG_RETRIEVE_START ALPM_EVENT_PKGDOWNLOAD_DONE -> ALPM_EVENT_PKG_RETRIEVE_DONE ALPM_EVENT_PKGDOWNLOAD_FAILED -> ALPM_EVENT_PKG_RETRIEVE_FAILED

these 3 have no equivalent:

ALPM_EVENT_RETRIEVE_START ALPM_EVENT_RETRIEVE_DONE ALPM_EVENT_RETRIEVE_FAILED

the only other with a FAILED status is ALPM_EVENT_DB_RETRIEVE_* and i don;t think it's the same thing. i could be wrong, but i have nothing to provide any useful feedback on this. maybe looking at the alpm source code itself might help if i can find it. that might well resolve the context issue anyway. but, back to the ALPM_EVENTs, when they come in they in turn fire a specific signal, so we need to know exactly how the old and new line up. that's where i'm at so far.

regarding alpm_option_set_dlcb (and all the other callback setters), compiling throws a warning on the second parameter which may or may not be a problem - i won;t be able to tell until i can compile and run. it could be that that parameter simply requires a cast, or it could be that the old callback functions have a different set of parameters than the new ones. either way, i'll know better when i can compile and run everything.

EDIT: i can confirm that the new callback functions do have a difference in their parameters - that context pointer is now required as the first parameter although the other parameters are the same. also, it appears that NULL can be used for the context pointer passed as the last parameter of the callback setter functions (note to @Thulinma). that's how pacman itself uses it. could be it's something planned for future use.

EDIT2: the README from the v6 code shows this as a change:

• ALPM_EVENT_RETRIEVE_ -> ALPM_EVENT_DB_RETRIEVE_

so, i guess that answers another of my questions.

anyway, i'm hitting the road in an hour or so, and won;t be able to get back into this until much later tonight at the earliest.

Thanks @jghodd. It's good to know there are potentially two people who can look at this. I've added a link to the package-query patch as a reference.

EDIT: I tried blindly replacing the text as per the package-query patch, but it wasn't enough. There were still errors with alpm_option_set_dlcb.

@Rhinoceros - thanks for the link. i'd do the patch, but the timing is really bad. i'm going to be out of town for the next 5 days and am not sure if i'll get a chance to dig into the source code. i'll see what i can do. meanwhile, if Thulinma has the time... (i'll keep an eye on any progress made). btw, package-query had similar issues and it's been fixed - might want to also take a look at how the maintainer fixed it. i know the calls to the set_ and get_architecture(s) methods were mended. that'd be a start. meanwhile, i have been able to build new working versions of yaourt, package-query, yay and paru. yaourt with a rebuild, package-query from the github fixes, and yay and paru from aur updates. package-query-1.12 should be available from the aur pretty soon.

Thanks @Thulinma, that's great. I've pinned your comment.

I do know C/C++ and would be more than happy to help patch kalu for pacman 6.0.

I even made a quick attempt a few hours ago (by working off of the existing 5.2 patch). Most of the changes are minor, there are two things that need to be fixed:

• the system architecture is no longer a single string setting, but a list.
• Several calls now have an extra parameter for "context", and it is unclear to me what this is for / what it does. I tried to find documentation on this part of libalpm, but it was hard to find after a cursory search.

Neither of these are huge issues. Both are just a matter of putting some time in to learn how they are meant to be used. Right now I don't have that time - but if nobody else picks this up in the next few days I'll probably end up doing it. If somebody else has more knowledge on the new context argument, that would cut most of the time I'd need to put into it and would thus help speed things up considerably. Any takers..?

@jghodd Yes, I agree. In the pinned comment, I've linked a working patch for alpm_octopi_utils that someone else submitted, which might be useful. I had a brief look at kalu-alpm.c to see if I could make similar changes, but unfortunately I don't understand C++ (the patch) nor C (kalu).

Looks like the issue is all contained to kalu-alpm.c. Patch, anyone?

pacman 6.0 :-(

This package is broken with pacman 6.0.0. Upstream hasn't responded for years now, but hopefully someone posts a patch somewhere (previously this was on GitHub), and I can apply the patch here.

EDIT: for reference, here is a working patch for alpm_octopi_utils, which might be useful. (It's in C++.)

EDIT2: package-query was also patched. Both kalu and package-query are written in C.

@noraj The previous version of plasma-workspace (5.20.5-2) provided notification-daemon, but the new version does not. You'll have to install a new package to provide this dependency instead, e.g. notification-daemon, but there are a few other options.