IRC channel logs

<Pellescours>with heavy fs operations (compiling rumpkernel), I feel that sometimes hurd (with rump as disk driver) is desync. Like a file is written/created but another process/command don’t see it/his modifications.

<Pellescours>sometimes I have some operation that fails and restarting the same operation works

<youpi>?

<youpi>I have never seen that

<youpi>and that shouldn't have to do with rumpdisk since ext2fs keeps things in cache, so it doesn't need to re-read from the disk

<Pellescours>I think it’s a specific case. I’m re-checking it but if this times works it related to a deletion and re-creation of files

<youpi>Pellescours: possibly your test case wants less-than-second file times precision?

<Pellescours>AH maybe

<Pellescours>I’m compiling packages with pacman, and it’s mainly (only?) the rumpkernel package that has this issue. I hit this issue a lot by re-constructing the package after it fails (I had to fix the PKGBUILD and relaunch the build)

<Pellescours>and FYI https://github.com/etienne02/cross-hurd/blob/archhurd/archhurd-repo/rumpkernel/PKGBUILD if you define some variables when compiling pci-userspace and librumpuser you are able to have every librairies in obj/destdir/usr/lib (this removes the needs of find that debian does)

<Pellescours>variables needed is DESTDIR and MKPICINSTALL=yes, the _SRC_tOP_ and MKUNPRIVED are because it tries to do a chown that fails on my case

<youpi>l'exemple était effectivement une blague, un rejet d'adhésion ce serait probablement plutôt par rapport à des personnes connues comme étant toxiques, par exemple

<youpi>EWIN

<Pellescours>x)

<damo22>you guys having inside jokes in french? :P

IRC channel logs

2025-03-29.log