Daniel LakelandI'm 4 hours into doing a #guix pull... its crashed multiple times I think either from out of memory or gnome going to sleep mid pull, along with all the debugging I had to do because I had some incompatible channel settings... after rebooting and killing X I think this time it'll work. then once the pull is over I can reconfigure the OS... all this is on a Chromebook like Celeron from 2018 with 4GB of RAM. Hopefully the reconfigure works and I can get some zram on here.
Note that on my kids 2021 era i5 laptops this completes in a few minutes. its some combination of small RAM and maybe mild mistakes on my part and slow processor that this takes so long.
I gave up on this, had my son change his password so I could work on his account on his laptop and managed to make it work!
i have a script that open a fhs compatibility container with guix shell and then codium running in regular session calls that script as its Julia binary... and it works!
roptat@dlakelan
Ouch… that's weird, because I can guix pull on a server that has 512 MB of RAM (and 1GB of swap). It takes about 10 minutes, and I need to stop all but network and SSH services for it to work, but it works…
Ouch… that's weird, because I can guix pull on a server that has 512 MB of RAM (and 1GB of swap). It takes about 10 minutes, and I need to stop all but network and SSH services for it to work, but it works…
@dlakelan
When it goes out of memory, the trick I use is to let it compute the derivation and show it (…-profile.drv I think), then stop the process and run "guix build /gnu/store/…-profile.drv", then "guix pull —commit" to the same commit, so it computes the same derivation and switch immediately to it. (The guix pull process does not free its memory after computing the derivation, so the build process will have less memory to work with otherwise)
When it goes out of memory, the trick I use is to let it compute the derivation and show it (…-profile.drv I think), then stop the process and run "guix build /gnu/store/…-profile.drv", then "guix pull —commit" to the same commit, so it computes the same derivation and switch immediately to it. (The guix pull process does not free its memory after computing the derivation, so the build process will have less memory to work with otherwise)
Khleedril@khleedril
This is probably what I will do. in the end I abandoned the whole project when I realized that there were no substitutes for some of the packages I was trying to install. Like librewolf. I'm pretty sure this little machine isnt gonna compile librewolf before the end of the universe. Similarly vscodium from nonguix
how do I combine a deploy and time machine?
@roptat
@roptat
Perfect, now here's the question. Suppose I look at the CI info and find that there were relatively few jobs 2 weeks ago so I'm likely to have substitutes. But there are thousands of commits since then. How do I find a commit from that timeframe? Also I need a relatively similar commit from nonguix
@khleedril
Perfect, now here's the question. Suppose I look at the CI info and find that there were relatively few jobs 2 weeks ago so I'm likely to have substitutes. But there are thousands of commits since then. How do I find a commit from that timeframe? Also I need a relatively similar commit from nonguix
@khleedril