Bug#1022173: Update (tested on bullseye, Libreoffice 1:7.0.4-4+deb11u1)
On Fri, 21 Oct 2022 21:42:45 +0200 Rene Engelhard wrote:
>True. But then they know or should know that stuff might become slow.
Actually not the stuff itself becomes slow all of a sudden, but the needed optimisations for this hardware often is removed from code, or efficient functions have been replaced by resource consuming eqivalents. Modern implementing of former hardware functions in software (like sound processing, video codec processing etc.) slows down clasic CPUs, while the dedicated hardware idles, since for everything the CPU is used. Hence some updated software runs slow on it (while other with same functionallity does not). A good example is the nouveau driver vs. proprietary nvidia driver. With the latter all GUI menus open immediately, without any delay, while with nouveau driver you'll have to wait sometimes more than 8 seconds for a response after click. So, if you chose your OS, the kernel, the software, the drivers and libraries carefully, you can see runing the so called deprecated hardware as fast as new one. It's an amazing experience, btw, this is just like you can see an old 1960s racing car catch up easily with any of the todays lower middle-sized class cars. But this is not the place for this basic discussion about old hardware I believe, so I won't go into detail. Just one last additional info: Libreoffice is one of the great programs running fast and fluent on old hardware, even in its recent versions.
>If I need to nudge them into that direction, I will do.
You'd better nudge nvidia to allow devs of xorg to update the nvidia kernel modules for all their older video cards, it is missing a tiny blob only, then people simply could upgrade to buster and bullseye. Its not the user not willing to update his system, but the hardware manufacuter. (Was quite tricky to force the last available proprietary nvidia driver for this notebook GPU run even on buster, but it was worth it.) Anyway, since I wasn't able to make it work on bullseye by now, the following test was run with nouveau driver instead.
Concerning the bug:
I've managed to run a test for Libreoffice on bullseye now, using a Live USB medium. Doesn't count whether the needed propriatary nvidia driver isn't available for bullseye, since for this testing the nouveau graphics driver is a basic replacement (not fit for everyday usage due to generally heavy delay in response, broken in dualhead and on resume after suspend on this device, anyway, this driver is the main issue keeping me from upgrading to bullseye completely.)
The interessting result is: I found the issue described by my original report (from backport to buster) present also in the non backported version on bullseye. Behaviour exactly as described above.
Version numbers as reported by Libreoffice GUI:
Version: 7.0.4.2
Build ID: 00(Build:2)
CPU threads: 1; OS: Linux 4.9; UI render: default; VCL: gtk3
Locale: de-DE (de_DE.UTF-8); UI: de-DE
Debian package version: 1:7.0.4-4+deb11u1
Calc: threaded
$ apt-cache policy libreoffice* | grep -v '(keine)' | grep -B1 Installiert
libreoffice-calc:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-base-core:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-core:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-common:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-draw:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-impress:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-l10n-de:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-style-colibre:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-writer:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-help-de:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-help-common:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-math:
Installiert: 1:7.0.4-4+deb11u1
libreoffice-gtk3:
Installiert: 1:7.0.4-4+deb11u1
ucf:
Installiert: 3.0043
libabw-0.1-1:
Installiert: 0.1.3-1
libc6:
Installiert: 2.31-13+deb11u2
libe-book-0.1-1:
Installiert: 0.1.3-2
libepubgen-0.1-1:
Installiert: 0.1.1-1
libetonyek-0.1-1:
Installiert: 0.1.9-4
libgcc-s1:
Installiert: 10.2.1-6
libicu67:
Installiert: 67.1-7
liblangtag1:
Installiert: 0.6.3-2
libmwaw-0.3-3:
Installiert: 0.3.17-1
libodfgen-0.1-1:
Installiert: 0.1.8-2
librevenge-0.0-0:
Installiert: 0.0.4-6+b1
libstaroffice-0.0-0:
Installiert: 0.0.7-1
libstdc++6:
Installiert: 10.2.1-6
libuno-cppu3:
Installiert: 1:7.0.4-4+deb11u1
libuno-cppuhelpergcc3-3:
Installiert: 1:7.0.4-4+deb11u1
libuno-sal3:
Installiert: 1:7.0.4-4+deb11u1
libuno-salhelpergcc3-3:
Installiert: 1:7.0.4-4+deb11u1
libwpd-0.10-10:
Installiert: 0.10.3-1
libwpg-0.3-3:
Installiert: 0.3.3-1
libwps-0.4-4:
Installiert: 0.4.12-1
libxml2:
Installiert: 2.9.10+dfsg-6.7
uno-libs-private:
Installiert: 1:7.0.4-4+deb11u1
zlib1g:
Installiert: 1:1.2.11.dfsg-2
$ lsb_release -a
No LSB modules are available.
Distributor ID: Debian
Description: Debian GNU/Linux 11 (bullseye)
Release: 11
Codename: bullseye
uname -r
4.9.0-279-antix.1-486-smp
$ lscpu
(same as above)
$ free
gesamt benutzt frei gemns. Puffer/Cache verfügbar
Speicher: 2061432 114436 509708 504928 1437288 1375304
Swap: 2097148 5936 2091212
This was a test using the preinstalled version as present in the Live USB device. I'll apply an apt-upgrade to this USB stick tomorrow and recheck. The installation-candidate for Libreoffice is: 1:7.0.4-4+deb11u4, which will be the next one I'm going to check out.
> And this report has a too high severity.
Sorry again for the trouble! As said before, I'm not experienced in bug reporting. I just follewed the instructions from https://www.debian.org/Bugs/Reporting and from the within there linked https://www.debian.org/Bugs/Developer#severities. The description
„important
Ein Fehler, der wesentliche Auswirkungen auf die Benutzbarkeit des Pakets hat, ohne es völlig unbrauchbar für jedermann zu machen.”
was what matched the problem best, I thought. I mean, an issue rendering the program mostly unusable without figuring tricky workarounds, due to excessive CPU load, is exactly what I'd put into this category. I'm fine with you having changed it to something else, while I still can't understand why you consider it only being
„minor
Ein Problem, das die Nützlichkeit des Pakets nicht beeinflusst, und das vermutlich sehr leicht zu beheben ist.”
since this issue keeps from using Libreoffice at all, if you don't know the proper workaround: It affects the complete system. (And whehter it is easy or difficult to fix I can't guess.) As said, I'm only a user, admitedly with some experience, but not a programmer, and probably my assessment of the issue was way to user-centric. Possibly you could clarify the description on this instructions site, so it is understood properly by everybody.
Maybe on modern multi kernel CPUs this happens without notice, since there is enough power present to simply mask this issue, so not slowing down other tasks noticibly when it happens, but only draining battery unnoticed. Just a guess.
>And was on a non-supported distro anymore. And for backports which doesn't belong here.
Yes, sorry, sorry, sorry.
If you want people to know about the proper procedure to be observed when reporting issues for backports, please add a sentence to https://www.debian.org/Bugs/Reporting instead of hiding this piece of information somewhere else. At least me had never expected I'd need to look into the link you've posted above before reporting a problem to the recent stable Libreoffice version, I simply didn't know it makes any difference for you whether it was backported or not. I was completely convinced the Libreoffive version itself is what counts.
Well, now I've reproduced the issue on a non backported bullseye finally, so I presume it's fine in this section now. Will check back with additional results once updated both the system and Libreoffice as said above.
Regards
Robin
Reply to: