IRC channel logs

<frogmo>Hello, the implementation of ext4 for the Hurd file system will aran it?

<frogmo>I find the current outdated ext2fs software on hurd very slow and outdated.

<youpi>how so "very slow" ?

<youpi>note that disk readahead is not implemented yet at the block layer, that one brings quite some slowness, which shouldn't be attributed to ext2fs

<frogmo>In the ext2fs tests that Hurd has against linux ext4, ext2fs is very obsolete.

<youpi>obsolete, that's a thing

<youpi>but what about speed?

<frogmo>Quite slow

<youpi>how so?

<youpi>how do you measure it?

<youpi>(again, see about readahead)

<frogmo>For installation tests

<youpi>(again, see about readahead)

<frogmo>Another one also at startup in the desktop environment, the system slows down, practically hanging.

<youpi>(again, see about readahead)

<frogmo>Also, when entering the desktop environment, the system slows down, practically hanging.

<youpi>(again, see about readahead)

<frogmo>I don't know why suddenly the page is out of order.

<youpi>which page ?

<frogmo>This hurd channel page

<frogmo>youpi It also said, when entering the desktop environment, the system slows down and crashes.

<youpi>possibly the desktop is just not portable and goes erratic when confronted with the hurd system calls behavior

<youpi>which desktop is that?

<frogmo>youpi Mate as Xfce, but it's been defective for a long time and they don't fix it.

<youpi>then help is probably welcome

<frogmo>youpi It would be nice if Hurd would move forward with the ext4 implementation, and fix those bugs to make it usable in the desktop environment.

<youpi>again, I don't think the problem lies in ext2fs

<youpi>ext2fs does work fine

<youpi>readahead is missing however, but that's not in ext2fs

<youpi>and desktops may be misbehaving because of subtle system calls behavior, but that's in glibc, not ext2fs

<frogmo>It is very slow ext2fs

<youpi>no it's not

<youpi>between a desktop environment and ext2fs there are countless layers, each of which can pose problem

<frogmo>And why don't they fix it in glibc ?

<youpi>help welcome

<youpi>maintainers don't magically have infinite available time

<frogmo>youpi Unfortunately I have no idea about programming

<youpi>you can learn

<frogmo>How?

<youpi>there are books, online courses, etc.

<gnu_srs>youpi: Something is definitely wrong: (on another image):apt-get install --reinstall dbus; pidof: stat failed for /proc/3074/root!

<gnu_srs>Can't start system message bus - /proc is not mounted ... failed!

<gnu_srs>And building latest hurd from source: mkdir: cannot create directory ‘/home’: File exists; make[1]: *** [override_dh_auto_install] Error 255;

<gnu_srs>/bin/fakeauth: Error 2 for child 2884; /bin/settrans: Error 2 for child 2883

<gnu_srs>Updated to the latest packages.

<gnu_srs>youpi: ?

<youpi>?

<gnu_srs>Maybe I should sen a bug report?

<gnu_srs>(18:53:17) gnu_srs: youpi: Something is definitely wrong: (on another image):apt-get install --reinstall dbus; pidof: stat failed for /proc/3074/root!

<gnu_srs>(18:53:17) gnu_srs: Can't start system message bus - /proc is not mounted ... failed!

<gnu_srs>(18:55:03) gnu_srs: And building latest hurd from source: mkdir: cannot create directory ‘/home’: File exists; make[1]: *** [override_dh_auto_install] Error 255;

<gnu_srs>(18:55:03) gnu_srs: /bin/fakeauth: Error 2 for child 2884; /bin/settrans: Error 2 for child 2883

<gnu_srs>(18:56:07) gnu_srs: Updated to the latest packages.

<youpi>I have no idea about that

<youpi>(except the aeternal advice: try to find a specific case that reproduces the issue, and then you can dig what exactly is producing the issue)

<gnu_srs>Just try to build hurd on an updated image?

<youpi>my box is up-to-date, and I can build the hurd fine

<gnu_srs>And apt-get install --reinstall dbus triggers the bug!

<youpi>it seems even dpkg-reconfigure dbus triggers it

<youpi>for the pidof message, a mere pidof call triggers it

<youpi>that should be easy to investigate then

<youpi>concerning the dbus start the message is probably coming from dbus itself

<youpi>trying to start it by hand would probably allow debugging it too

IRC channel logs

2022-01-24.log