First, I’d like to thank all the people working on CachyOS. It feels great—like the “new Ubuntu” back in 2005.
Installation is usually very fast thanks to pre-compiled packages. But recently, I installed Citrix Workspace (paru -Syu icaclient) and ended up with over four hours of compiling time at 100% CPU usage.
For this program, there seems to be a Flatpak available, which would shorten the installation time drastically. It would be great to get a hint before entering a multi-hour compilation session.
Nevertheless, I had a similar problem when I needed to compile a custom kernel due to a regression.
Is there any chance to get a time estimation before compilation starts?
So used a foreign build script to manually compile the software from source.
Well, yeah, but it is not the same thing.
Its not just pre-compiled .. it is a flatpak, which is containerized - requiring the flatpak framework, its own flatpak dependencies, and other considerations.
The hint is that you applied a command which does certain things, using an AUR-helper, and then were presented with exactly what was going to happen (PKGBUILD is shown by paru by default) to which you accepted everything.
Though this does not specifically include output of an estimated build time.
I do not know if such a thing would be at all possible.
Certainly not without at least downloading the sources and such first - considering these build scripts are for foreign things so the software could be anything held anywhere.
Or some kind of of huge database of how big each AUR package is after having built it? It cannot be simply an index of static times because each build environment is different with different resources.
In any case as of now it is not part of any existing tool that I know of.
Well, yeah, but it is not the same thing.
Its not just pre-compiled .. it is a flatpak, which is containerized - requiring the flatpak framework, its own flatpak dependencies, and other considerations.
Yes, I understand that it might be a little bit against the philosophy of the optimized os. Nevertheless, sometimes the priority of special software is, that it just runs.
I do not know if such a thing would be at all possible.
Certainly not without at least downloading the sources and such first - considering these build scripts are for foreign things so the software could be anything held anywhere.
Or some kind of of huge database of how big each AUR package is after having built it? It cannot be simply an index of static times because each build environment is different with different resources.
In any case as of now it is not part of any existing tool that I know of.
Maybe I am thinking too easy because I am not a developer. The compiling time can be estimated after compiling. If this is combined with the CPU information (lscpu) and uploaded, we would get a dataset. It seems that Geekbench has the performance scores of almost all CPUs available. So, for the next user, the compiling time from the dataset multiplied by the Geekbench CPU score should result in a rough estimate of the compiling time.
A key thing to keep in mind is that the AUR is completely independent of Cachy and not officially supported in any way. As a convenience, you have a helper available to access it if you choose.
Ultimately it’s just a community resource of scripts to download and install/build software that isn’t in the official repos, with no guarantees attached other than your ability to inspect what you get.
Who on earth wants to
switch to a new and untrusted aur helper with this feature
publicly share telemetry of this nature
receive no benefit because whatever small chunk of people who have switched are installing different software anyway.