<htsr>I'm reinstalling GuixSD with the qemu image but I don't understand the note at starup. It says "Expand the root partition to fill the space available by 0) deleting and recreating the partition with fsdisk, [...]" but the / partition is the first. So I need to move the second and then delete and recreate? Or the second is not important? (It seems to containt the boot)
<vagrantc>gah. i just tried to configure a desktop for the c201, and it rendered the system ... not unbootable, but it was impossible to log in ... the logins on the ttys failed and slim didn't successfully start
<vagrantc>and depthcharge does not make it particularly easy to access prior system generations...
<leungbk>how can i apply a .patch file to a .el file i am packaging? when i try applying my patch, i am told in the build log that my thing is not a .tar file.
<sneek>leungbk, jackhill says: perhaps instead of a whole separate package a separate output would do? However, does having the elisp files (and compiled elisp files) available force the user to install emacs? My untested assumption would be that it would not (just like having info files doesn't force a user to install an info reader). Of course emacs will be required to build, but if a user is able to get a
<leungbk>oh, does this mean i should use (replace 'install [...])?
<nckx>leungbk: If you're talking about jackhill's sneek message: not per se. It probably means just leave the elisp files where they are (they're not large) or use (outputs "out" "foo") if you really want to separate them.
<rvgn>I started to wonder why GNU emacs adopted so many feature that are very far from it's initial objective. Because, GNU Project follows Unix Philosophy of "Make each program do one thing well. To do a new job, build afresh rather than complicate old programs by adding new "features". It is confusing to see GNU emacs (a GNU Package) deviated from GNU'
<vagrantc>herd: exception caught while executing 'start' on service 'file-system-/sys/fs/cgroup/perf_event'
<vagrantc>only happens when i try to enable a desktop system... but it breaks console logins too ... i'm hoping it's just a matter of enabling CONFIG_CGROUP_PERF in the kernel ... but that seems to depend on some other options
<rekado>apteryx: I found that the pypi importer has become a little less useful over time. I’ve encountered many packages where no depenencies could be detected at all.
<efraim>I finally realized that shepherd is at /run/current-system/profile/bin/shepherd and I don't need to install it locally to have my own shepherd services
<efraim>perhaps if it can't find a requrements.txt it should spit out a message to check setup.py, but I think it'd become obvious when it fails to build
<efraim>perhaps have it also parse test-requrements.txt if it does find requirements.txt and add it to native-inputs
<apteryx>our current scheme is to look into the wheel metadata.json for a "run_requires" list, else fallback on a */.egg-info/requires.txt file, which resembles a INI.
<apteryx>(and we used to also look in requirements.txt before requires.txt, but I've just stripped this out)
<apteryx>my understanding is that if a package has a setup.py, it'll have the metadata.json or requires.txt populated. If it doesn't have that, well, it's not on PyPI and thus would need a careful look at how they did things anyway.
<apteryx>requirements.txt is not warranteed to be there, and if it is, it might contains stuff we don't want (typically used for freezing a development environment, including extra tools).