IRC channel logs

2024-10-15.log

back to list of logs

<damo22>youpi: where is the buildd installed? can i see a list of broken packages?
<damo22>this looks like a good place to start v
<damo22> https://people.debian.org/~sthibault/hurd-i386/graph-top.txt
<damo22>so i fixed python-psutil but i dont know how to implement all of the api
<damo22>ie it builds now
<youpi>damo22: https://www.debian.org/ports/hurd/hurd-devel-debian
<janneke>is there a README or something for hurd/web.git? i'm getting
<janneke>ymlfront: failed to use YAML::Syck
<janneke>running ./render_locally
<youpi>perhaps you need libyaml-syck-perl
<janneke>youpi: that's probably it; i now tried "guix shell ikiwiki perl-yaml perl-yaml-libyaml", but it seems guix doesn't have have anything "syck"
<bool>hello! in the project ideas page, "fixing programs using PATH_MAX et al Unconditionally", there is a link to a list of build failures (http://unstable.buildd.net/buildd/hurd-i386_Failed.html) however this link is either dead or I can't access it for some reason. any mirrors or other ways to access it? i'd like to contribute a little.
<youpi>bool: which page do you mean exactly?
<bool> https://www.gnu.org/software/hurd/community/gsoc/project_ideas/maxpath.html
<damo22>bool: this page has all the links you probably need https://www.debian.org/ports/hurd/hurd-devel-debian
<youpi>yes, I have now added it on that page
<damo22>hurd doesnt have struct sysinfo from linux it seems
<damo22>but it has sysinfo.h ?
<youpi>yes, sysinfo.h is not only about the sysinfo structure
<damo22>do we want to have it?
<bool>damo22: thank you! this seems to be exactly what i needed. and thanks youpi for the speedy fix on the page. im looking forward to helping out in anyway i can
<youpi>it can be useful
<damo22>im trying to figure out what is missing to implement python-psutil
<damo22>can i paste 10 lines?
<youpi>you can pastebin them ;)
<youpi> https://paste.debian.net/
<damo22>eg this https://paste.debian.net/plain/1332340
<youpi>affinity_get/set will be trivial for now (ENOSYS) since we don't really have cpu binding as is expected by cpu_affinity_get/set
<youpi>ioprio_get/set we don't have it for now, so ENOSYS too
<youpi>for net_if, we don't have the interface either
<youpi>for _users, I guess the linux version can be reused since unix users are a userland thing, not related to the system
<damo22>ok
<youpi>disk_partitions is a different beast, we don't really expose the partitions in some way for now
<damo22>what about mtab
<youpi>that's the mounted partitions
<youpi>it's different from linux' /proc/partitions
<damo22>i see
<damo22>the linux implementation of psutil disk is mtab
<youpi>oh?
<youpi>I thought it was about available partitions, not mounted partitions
<youpi>then our mtab should be fine
<damo22>// Return disk mounted partitions as a list of tuples including device,
<damo22>// mount point and filesystem type.
<damo22>PyObject *
<damo22>psutil_disk_partitions(PyObject *self, PyObject *args) {
<damo22>ok
<youpi>ok
<damo22>i will add arch/gnu then
<damo22>aix bsd freebsd linux netbsd openbsd osx solaris windows
<youpi>thanks!
<damo22>should i push it upstream?
<damo22>as PR
<youpi>of course
<damo22>ok
<youpi>there is no real need to go through debian actually
<youpi>we can push to upstream and upload the patch to debian's unreleased
<youpi>no need to bother the debian maintainer, as that'll land eventually anyway
<damo22>ok cool
<damo22>i see many python3 packages blocked on python-psutil
<youpi>yes
<damo22>ok i will fix it now
<damo22>what can i do about free memory reporting
<damo22>eg totalram, freeram etc
<youpi>you can see how vmstat etc. get them
<damo22>ok
<youpi>you could also as well just read it from /proc like linux does
<youpi>thus sharing the implementation rather than writing another one
<damo22>if proc has them id rather reuse it
<damo22>so i can try compiling the linux versions and stub / implement the broken ones
<youpi>note: concerning llvm-toolchain-19, the fix is already pending upload
<youpi>and concerning java, for me that's a lost cause, better just disable the dep in the packages that ask for it, they usually already have the option to do so since some other ports don't have java either
<janneke>ACTION adds perl-text-markdown, perl-yaml-syck cpan packages to guix
<janneke>and successfully builds the website doing
<janneke>(the ikiwiki package's check phase is currently broken...)
<janneke>guix shell --without-tests=ikiwiki ikiwiki perl perl-text-markdown perl-search-xapian perl-yaml-syck texinfo
<janneke>./render_locally
<etno>bool`, I happen to be chasing PATH_MAX as well; on netsurf currently.
<etno>I just hope that people upstream are not going to simply reject the patch...
<damo22>Successfully built psutil-5.9.8-cp312-cp312-hurd_i686.whl
<damo22>NotImplementedError: /proc/diskstats nor /sys/block filesystem are available on this system
<damo22>the rest kind of works i think
<youpi>good :)
<youpi>you can send me a patch for upload in unreleased
<damo22>email to private or on bug-hurd?
<damo22>i refactored the patch and it applies almost clean to both upstream and debian
<youpi>better on bug-hurd
<damo22>ok
<damo22>Python 3.12.7 (main, Oct 3 2024, 15:15:22) [GCC 14.2.0] on gnu0
<damo22>>>> import psutil
<damo22>>>>
<damo22>\o/
<damo22>gtg to sleep, but i did send in the patch to our list, might hold off sending PR to upstream just yet
<gnucode>hello friends!
<gnucode>janneke: So you are wanting to edit the wiki from guix system right?
<gnucode>If you get it working....may I ask you to contribute a guix.scm for the Hurd web repositories? I have only ever been able to edit the Hurd wiki from Debian. It would be nice to let potential contributers edit the wiki, with a guix.scm
<gnucode>sneek later tell bool I personally find it really easy to edit the Hurd wiki. I typically look through changes that I can make to the open issues pages.
<sneek>Okay.
<gnucode>seek later teel bool or I grep through the irc logs to find cool tibbits to add to various pages.
<gnucode>seek later tell bool or I grep through the irc logs to find cool
<gnucode> tibbits to add to various pages.
<gnucode>sneek later tell bool or I grep through the irc logs to find cool tibbits to add to various pages.
<sneek>Okay.
<gnucode>damo22: what is python-psutil? Why do so many packages depend on it?
<etno>gnucode: I think that it is the prefered way to access to process information from within python apps
<gnucode>ok. cool.
<youpi>gnucode: python-psutils is used by python-greenlet, which is used by wxwidgets, which is used by a lot of packages
<gnucode>etno: are you tidying up your netsurf patches? I have similar ones, but in some of the files I just define PATH_MAX.
<gnucode>weird...X seemed to lock up on me again.
<gnucode>C-A-Del
<gnucode>restart x
<gnucode>maybe it was just temporarily froze.
<gnucode>I've noticed that if it starts to lag...not respond to mouse movements...then I have about 5-10 seconds to press C-A-Del
<gnucode>otherwise it's too late.
<gnucode>youpi: do we use the checkperms translator? It's documented on the wiki, but i don't see it in /hurd/
<gnucode>janneke: also, the hurd has some upcoming fun stuff. game jam, hackathon, and a spooky halloween party. You're welcome to come!
<etno>I had put this patch on hold for a moment, because I found azert's comment rather pragmatic and I wasn't sure that strict standard conformance was convincing upstream. I resumed now; I remove PATH_MAX references by using dynamic allocation, as suggested by Samuel in his article.
<youpi>gnucode: as mentioned on the wiki it's in the checkperm-deferred-authorization branch
<gnucode>I assume that it is not super stable at the moment, since it's not in /hurd
<youpi>not necessarily
<youpi>personnally, I didn't even know about it
<youpi>so it's very possible that it was just not pushed by its author
<youpi>that's the usual problem: once one has finished one's toy, going further is less easy to find motivation for
<gnucode>Well, I can certainly try to compile it and see what happens.
<gnucode>BTW, I'm working on documentating all the undocumented translators that I see in /hurd
<gnucode>so I'm wanting to write a good example on the remap page translator page. If remap doesn't exist on the wiki yet, then it's because I haven't pushed it.
<gnucode>anyway, my thought is, it would be really cool to have remap set up such that I can test lwip.
<gnucode>I've got the eth-multiplexor set up.
<gnucode>pfinet uses /dev/eth0m/1
<gnucode>now I just need to set up lwip to use /dev/eth0m/2
<gnucode>fsysopts /servers/socket/26
<youpi> https://archive.fosdem.org/2017/schedule/event/microkernel_virtualization_on_hurd/attachments/slides/1593/export/events/attachments/microkernel_virtualization_on_hurd/slides/1593/virtualization_on_the_hurd.pdf
<gnucode>shows me my IP address, netmask etc.
<gnucode>I'll take a look. thanks.
<youpi>there's another example on https://lists.debian.org/debian-hurd/2016/08/msg00016.html
<gnucode>thanks!
<youpi>an interesting example can be to remap /bin/sh or /usr/bin/python to something else
<youpi>for a user to use something else than what is installed on the system
<youpi>without having to modify all their scripts
<gnucode>"an interesting example" I guess I don't follow that... remap /usr/bin/python3 ~/bin/my-python3-compiled-with-experimental-feature -- my-python-script.py
<gnucode>something like that?
<gnucode>I just thought of something cool....
<gnucode>technically the hurd has read support for ext3!
<gnucode>there is an existing translator right?
<gnucode>I don't know how stable that read support is...
<youpi>gnucode: for just a python script that wouldn't be useful, since you could just call the interpreter explicitly
<youpi>but remap /usr/bin/python3 ~/bin/my-python3 -- ./configure
<youpi>yes
<youpi>in case ./configure doesn't provide a way to select the python interpreter to be used to detect libraries etc.
<youpi>another example is remapping /servers/socket/2 and 26
<youpi>to isolate a program in terms of tcp/ip access
<youpi>to e.g. make it use a vpn
<youpi>without affecting other programs
<gnucode>that's another example that I should write down.
<janneke>gnucode: re edit wiki: err, no i was preparing to update the static markdown pages for Guix status
<janneke>and thanks for the invites :)
<janneke>although i'll probably not make it to these, i'll see
<gnucode>janneke: you are welcome!
<gnucode>settrans -c lwip/servers/socket/2 /hurd/lwip -i /dev/eth0m/1 -a ADDRESS \
<gnucode>> -m ADDRESS -g ADDRESS -4 ~/lwip/servers/socket/2 -6 ~/lwip/servers/socket/26
<gnucode>settrans -c lwip/servers/socket/2 /hurd/lwip -i /dev/eth0m/1 -a
<gnucode> ADDRESS \ [12:45]
<gnucode><gnucode> >settrans -c ~lwip/servers/socket/2 -m ADDRESS -g ADDRESS -4 ~/lwip/servers/socket/2 -6 ~/lwip/servers/socket/26
<gnucode>remap /servers/socket/2 ~/lwip/servers/socket/2 -- ping -c 3 gnu.org
<gnucode>that seemed to work.
<gnucode>though it would be nice, if I didn't have to specify -a -g -m everytime i reboot.
<gnucode>gotta go to work.
<youpi>sneek later tell gnucode you shouldn't have to pass -a -g -m, since you didn't use the -a option of settrans. Doesn't gettrans on the node show you that it kept the parameters?
<sneek>Got it.
<gnucode>youpi: that's actually cool that I don't have to pass the -g -m and --address of lwip. That makes it soo much easier!
<sneek>Welcome back gnucode, you have 1 message!
<sneek>gnucode, youpi says: you shouldn't have to pass -a -g -m, since you didn't use the -a option of settrans. Doesn't gettrans on the node show you that it kept the parameters?
<gnucode>I'm not actually certain what getttrans does. I've heard of settrans and showtrans
<youpi>it reads what's written on the disk
<youpi>thus what survives a reboot
<gnucode>I've also never seen settrans --chroot
<youpi>it's very rarely useful
<gnucode>youpi: I don't have a gettrans command...
<youpi>showtrans , sorry
<gnucode>gotcha
<gnucode>hmmmm remap /usr/bin /dev/null -- uname -a -> outputs the uname command
<gnucode>I thought remap /usr/bin to /dev/null should mean that I cannot execute any commands...
<youpi>damo22: erf, python-greenlet also needs some porting apparently