Home » Developing U++ » Releasing U++ » Archlinux AUR (Considering adopting U++ packages)
Re: Archlinux AUR [message #47928 is a reply to message #47912] |
Mon, 24 April 2017 21:21   |
|
Hi Eremiell!
Welcome to the U++ forum
Eremiell wrote on Sun, 23 April 2017 03:25I found U++ only recently (being many years C++ developer and tutor) and decided to give it a try. Sadly, reporting issues with AUR packages for nightly and git builds caused them to be orphaned. (No reaction from stable build yet.)
Maintainer of the AUR packages speaking here I'm sorry I didn't reply right away when you flagged the package in AUR. Unfortunately, I'm rather busy lately and I must admit I didn't take proper care about the packages. Feel free to contact me directly if you're interested.
Eremiell wrote on Sun, 23 April 2017 03:25I'm now considering for longer time to start maintaining some AUR packages, as I dwell into PKGBUILDs nearly daily, so I could as well put this skill up for some good.
So my first question would be if that's OK with you?
I'd be more than happy to transfer those packages into good hands of another Arch user who will have more time to keep them up to date.
Eremiell wrote on Sun, 23 April 2017 03:25If so, I'd like to ask about several things that could be done on U++ side, that would simplify the packaging a lot.
1) the builds as are now are using "uppbox/lpbuild2/mkfile". Could this be included into nightlies (and stables), so it doesn't have to be pulled from github, which causes potential version mismatch? Or could this probably be completely replaced by another build mechanism present? (I'm asking the later one simply because I don't know. If so, please enlighten me.)
That file is a wonderful thing, but it is not really needed for the basic packaging. Less then a year ago, Amrein did a many fixes in the generated makefile that is included in the tarball when he was working on RPM packages. It should be fairly easy to use the regular makefile for Arch packages as well.
Points 2) and 3) were answered by Mirek already And I agree with 4) as well.
Eremiell wrote on Sun, 23 April 2017 03:25Also the "#define DDUMP(x) @" in git is nasty, as it obviously emits errors, but they're sparse in a long build log that most people probably don't watch and while the build ends in error, the upp and umk packages still somehow "get done", only theide fails hard. But fixed that one with a simple sed. Just saying. Hope you don't plan to extend this strategy around. I think that it is actually pretty good idea. The only important part that is missing is that nightly builds should not be published if they don't pass the tests. Mirek do you think it would be possible to arrange that (if it's not already)?
Best regards,
Honza
|
|
|
Goto Forum:
Current Time: Tue May 13 11:12:52 CEST 2025
Total time taken to generate the page: 0.03460 seconds
|