On 6/22/21 9:49 AM, alad via aur-general wrote:
On 18/06/2021 19:59, Caleb Maclennan via aur-general wrote:
so let’s start the discussion period. This message is a short public service announcement reminding all current TUs that today is the halfway point of the discussion period on my application. I know packagers tend to be the easily distracted types
On 2021-06-11 03:16, Archange wrote: that wander off following the latest shinny thing wherever it leads — it takes one to know one. I also know the author of *the* most popular AUR package (not to mention co-author of the second most popular) just threw his hat in the ring too, but I don't want you to loose your chance to hold my toes over the fire.
Since I don't want to divert all attention to the audacity controversy, let me note my appreciation of your recent activity on various Arch community channels (e.g. IRC and mailing lists here). Whatever the outcome of this application, I hope you keep that activity.
Seconded. :)
Regarding your AUR packages, I guess we can't expect a full review of all 500. So let's talk about a few "random samples" and the packages you want to add to community.
* github-desktop
- That DLAGENTS logic looks strange. I guess makepkg doesn't support source links with both branch and tag?
This appears to be a homebrew implementation of the following rejected makepkg feature request: https://wiki.archlinux.org/title/User:Apg#makepkg:_shallow_git_clones
- Googling "Xvfb yarn" gave me this rather amusing issue: https://github.com/cypress-io/xvfb/issues/98
xvfb-run is probably easier here...
* moonscript
- I'd consider adding write_moon{,c} to the git repository, instead of putting long one-liners in the PKGBUILD. - You can use parameter substitution instead of sed. For example, ${pkgver/_/-}
I'll add that _lua_ver="$(lua -e 'print(_VERSION)' | cut -d ' ' -f 2)" gets evaluated before /usr/bin/lua is installed, and therefore evaluates to the empty string every time the PKGBUILD is sourced, while logging: PKGBUILD: line 21: lua: command not found to stderr. Due to an internal implementation detail of makepkg, the package() function is run after sourcing the PKGBUILD again -- this time after lua is installed. Luckily, the variable is only *used* during package(), so the resulting .pkg.tar.zst isn't created with the wrong content... but then it should be defined during package() too. As it is, it will simply produce stderr every time the PKGBUILD is sourced by default, which can break tooling. When I tested building it with makechrootpkg to test whether/where any wonkiness resulted, I noticed along the way that it reports: Missing dependencies for moonscript 0.5.0-1: lpeg >= 0.10, ~= 0.11 (not installed) alt-getopt >= 0.7 (not installed) luafilesystem >= 1.5 (not installed) then proceeds to download all 3 from https://luarocks.org/, compile them locally, and I think do nothing with them. Should be updated to depend on all three from the official repos, I guess. I think they're there with the right versions? I know nothing about lua... -- Eli Schwartz Bug Wrangler and Trusted User