IRC channel logs

<Darelelve>hwpplayer1: As a newcomer, I wish GNU/Hurd had more complete documentation on libraries, as well as more examples that were simple enough to show specific aspects of a given library.

<hwpplayer1>I see Darelelve

<hwpplayer1>Thanks

<hwpplayer1>sleep

<hwpplayer1>sleep time

<solid_black>afternoon!

<solid_black>why does lenovo.com of all places have an explanation of the Hurd

<solid_black> https://www.lenovo.com/us/en/glossary/gnu-hurd/

<youpi>wat :D

<solid_black>hi youpi \o/

<youpi>heya solid_black

<solid_black>i'm totally setting up a new hurd vm on this laptop

<solid_black>and a cross build environment

<solid_black>time to remember how to build glibc

<solid_black>then maybe it'll lead somewhere

<solid_black>one no longer has to use -M q35?

<youpi>apparently, depends how much memory you give it, some devices only support < 32b memory

<solid_black>less than thirty-two bytes? :D

<youpi>b, not B ;)

<sam_>I noticed the same thing and found it bizarre

<youpi>what I find bizarre is that the BSD drivers don't seem to be caring about the issue

<youpi>which, yes, is real

<youpi>just like the old 16MiB for DMA

<solid_black>could someone point me to what I need to put into /etc/apt to be getting recent improvements (& recent bugs) for Mach/Hurd/glibc but not all packages in the world?

<Darelelve>Hello, youpi ! Could you tell me how important is the mechanism for opening the store when an attempt is made to access it? The thing is, trivfs has a large number of hooks (getroot_hook, check_open_hook, etc.), and the only place where their code can be applied is in netfs_check_open_permissions, but that doesn't seem very neat, in my opinion. Opening the store immediately (as well as its sections, if any) simplifies the code...

<youpi>Darelelve: for e.g. a cdrom, we need to open as requested

<Darelelve>youpi: Got it. Thanks. Also, is it better for me to change the storeio code itself or write a new storeio from scratch based on the existing code?

<azeem>solid_black: I just run something like "sudo apt install libc0.3 hurd gnumach-image-1.8-amd64-up gnumach-image-1.8-amd64-up-dbg"

<youpi>Darelelve: better just change the existing storeio code

<Darelelve>youpi: Understood, thanks again!

<solid_black>was gnumach's debug info package fixed?

<solid_black>I investigated it a while back and figured out that it was simply missing a build-id, which is why debian's tooling was not picking it up

<solid_black>are error_t and kern_return_t not compatible in the eyes of GCC now?

<solid_black>oh that's not it at all, rather GCC is unhappy about the mismatch between size_t and mach_msg_type_number_t

<solid_black>and that rings a bell, apparently this code was never before compiled for a 64-bit system :D

<solid_black>turns out, I did have much more 9pfs work lying around on 'hurdle', and it does work so much better

<sam_>there's a simple fix to get latest improvements for things, you know..

<jab>p9fs on the Hurd would be pretty cool to see.

IRC channel logs

2026-06-01.log