View previous topic :: View next topic |
Author |
Message |
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Tue Oct 28, 2008 2:45 am Post subject: |
|
|
audiodef,
Use ccache -- your work will be saved, so if you have to re-emerge a package it will go (relatively) quickly.
As for collision-protect, I don't think it's very useful. Large packages like GNOME, OpenOffice, KDE, etc. move files around between subpackages/dependencies, and this trips collision-protect. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
NathanZachary Moderator
Joined: 30 Jan 2007 Posts: 2605
|
Posted: Tue Oct 28, 2008 3:35 am Post subject: |
|
|
I personally don't like finding an application that I want is not available in the standard tree. I am not keen on using overlays, and for a simple package to sit in "maintainer wanted" is frustrating. Maybe one day when I am comfortable at writing and maintaining ebuilds, I will be able to help out. I also don't understand why certain things are more complicated/seemingly impossible to get working in Gentoo but not in other distros. The two that come to mind are the Firefox slowness when viewing a page with the background CSS, and ntfs-3g automounting in KDE. Other than those, I really like all the aspects of Gentoo. I don't even mind compilation time. _________________ “Truth, like infinity, is to be forever approached but never reached.” --Jean Ayres (1972)
---avatar cropped from =AimanStudio--- |
|
Back to top |
|
|
few Guru
Joined: 03 Mar 2008 Posts: 448
|
Posted: Tue Oct 28, 2008 7:21 am Post subject: |
|
|
audiodef wrote: | OMFG! Hours and hours compiling OpenOffice (downgrading from the buggy 3.0 version) and then it tells me it can't complete due to file collisions. Unmerged, figuring that'd solve the problem and ran it again, but NO. Collisions. Just deleted the effin' /usr/lib/openoffice dir.
Really need to get on the ball, folks, so that users are warned BEFOREHAND whether something WILL fail.
Really hate that about Portravage. |
Use FEATURES="buildpkg" and you will save a lot of time. (the package is even created if collisions occur) |
|
Back to top |
|
|
drescherjm Advocate
Joined: 05 Jun 2004 Posts: 2790 Location: Pittsburgh, PA, USA
|
Posted: Tue Oct 28, 2008 1:01 pm Post subject: |
|
|
Quote: | Use ccache -- your work will be saved, so if you have to re-emerge a package it will go (relatively) quickly. |
Agreed. ccache also helps significantly when you you upgrade a package and most of the source code has not changed. To me this should be on by default. _________________ John
My gentoo overlay
Instructons for overlay |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Tue Oct 28, 2008 1:04 pm Post subject: |
|
|
Similarly, a change in a USE flag rarely affects more than a small portion of the code. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
Yamakuzure Advocate
Joined: 21 Jun 2006 Posts: 2285 Location: Adendorf, Germany
|
Posted: Tue Oct 28, 2008 1:41 pm Post subject: |
|
|
OpenOffice 3 is buggy? That explains why I got it stable on one machine, and on another (I sync'ed there one day later) it is still (or aigain) marked as unstable...
...but I haven't had a single problem with ooo3 on this machine, and I am working quite alot with it... But then this could be sheer luck, of course. _________________ Important German:- "Aha" - German reaction to pretend that you are really interested while giving no f*ck.
- "Tja" - German reaction to the apocalypse, nuclear war, an alien invasion or no bread in the house.
|
|
Back to top |
|
|
Geralt Apprentice
Joined: 14 Jun 2008 Posts: 150
|
Posted: Tue Oct 28, 2008 1:43 pm Post subject: |
|
|
Yamakuzure wrote: | OpenOffice 3 is buggy? That explains why I got it stable on one machine, and on another (I sync'ed there one day later) it is still (or aigain) marked as unstable...
...but I haven't had a single problem with ooo3 on this machine, and I am working quite alot with it... But then this could be sheer luck, of course. |
That depends, a package is sometimes marked as unstable because nobody could test it, yet, on such a machine. |
|
Back to top |
|
|
drescherjm Advocate
Joined: 05 Jun 2004 Posts: 2790 Location: Pittsburgh, PA, USA
|
Posted: Tue Oct 28, 2008 1:51 pm Post subject: |
|
|
Since oo3 is new in my opinion it should still be in testing. And being in testing for me does not mean that an application is buggy. Just like being in stable does not mean the application is bug free. _________________ John
My gentoo overlay
Instructons for overlay |
|
Back to top |
|
|
audiodef Watchman
Joined: 06 Jul 2005 Posts: 6640 Location: The soundosphere
|
|
Back to top |
|
|
maltheus Tux's lil' helper
Joined: 14 Aug 2004 Posts: 125 Location: Colorado Springs
|
Posted: Tue Oct 28, 2008 2:54 pm Post subject: |
|
|
Blocked packages really tick me off. If the solution usually involves unmerging or emerging in a different order, why can't portage handle that? And although the solution is usually to just unmerge the blocker, sometimes that will also break your system so I'm always forced to check the forums first. Surely the proper instructions could be encoded into the ebuild somehow. |
|
Back to top |
|
|
audiodef Watchman
Joined: 06 Jul 2005 Posts: 6640 Location: The soundosphere
|
|
Back to top |
|
|
drescherjm Advocate
Joined: 05 Jun 2004 Posts: 2790 Location: Pittsburgh, PA, USA
|
Posted: Tue Oct 28, 2008 3:22 pm Post subject: |
|
|
Quote: | Blocked packages really tick me off. If the solution usually involves unmerging or emerging in a different order, why can't portage handle that? |
portage-2.2 handles these much better than before but can still get into this situation. _________________ John
My gentoo overlay
Instructons for overlay |
|
Back to top |
|
|
Geralt Apprentice
Joined: 14 Jun 2008 Posts: 150
|
|
Back to top |
|
|
torontolimo n00b
Joined: 28 Oct 2008 Posts: 3
|
|
Back to top |
|
|
viralex Apprentice
Joined: 24 Apr 2008 Posts: 237 Location: Viareggio (Lu,Italy)
|
Posted: Tue Oct 28, 2008 6:22 pm Post subject: |
|
|
I know that the label 'amd64' is right, but I'd like to see 'x86_64' instead!
I'm on a intel quad! |
|
Back to top |
|
|
yngwin Retired Dev
Joined: 19 Dec 2002 Posts: 4572 Location: Suzhou, China
|
Posted: Wed Oct 29, 2008 12:04 am Post subject: |
|
|
viralex wrote: | I know that the label 'amd64' is right, but I'd like to see 'x86_64' instead! |
Not going to happen. Renaming an arch would be a painful process, and it's just not worth the trouble. Besides, we use x86 and i686 also for AMD Durons and Athlons, so why should there be a problem with amd64? _________________ "Those who deny freedom to others deserve it not for themselves." - Abraham Lincoln
Free Culture | Defective by Design | EFF |
|
Back to top |
|
|
jonnevers Veteran
Joined: 02 Jan 2003 Posts: 1594 Location: Gentoo64 land
|
Posted: Wed Oct 29, 2008 2:10 am Post subject: |
|
|
torontolimo wrote: | The time it takes to compile a package... |
gentoo is a source based distro. <laughing at complaining about compile times/>
viralex wrote: | I'm on a intel quad! |
well then, talk to Intel. they were the ones playing games way back when. amd took the initiative and delivered the first 64bit compatible microprocessors (for a consumer chips). Intel only joined the party when they realized which way the inevitable tide was flowing. |
|
Back to top |
|
|
admax88 n00b
Joined: 23 Aug 2008 Posts: 21
|
Posted: Wed Oct 29, 2008 3:02 am Post subject: |
|
|
I love gentoo, I think its the best distribution out there.
The one thing that I see could use improvement is package specific use flag documentation. It would be nice to know exactly what effect a use flag has on a specific package before compiling. |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Wed Oct 29, 2008 4:29 am Post subject: |
|
|
admax88 wrote: | The one thing that I see could use improvement is package specific use flag documentation. It would be nice to know exactly what effect a use flag has on a specific package before compiling. |
Others (and I) have mentioned this in this thread, but it's worth repeating. Let me give a specific example I encountered recently in ffmpeg; from the ebuild:
Code: | RDEPEND="vhook? ( >=media-libs/imlib2-1.4.0 >=media-libs/freetype-2 )
sdl? ( >=media-libs/libsdl-1.2.10 )
encode? (
aac? ( media-libs/faac )
mp3? ( media-sound/lame )
vorbis? ( media-libs/libvorbis media-libs/libogg )
theora? ( media-libs/libtheora media-libs/libogg )
x264? ( >=media-libs/x264-0.0.20081006 )
xvid? ( >=media-libs/xvid-1.1.0 ) )
aac? ( media-libs/faad2 )
zlib? ( sys-libs/zlib )
ieee1394? ( media-libs/libdc1394
sys-libs/libraw1394 )
dirac? ( media-video/dirac )
gsm? ( >=media-sound/gsm-1.0.12-r1 )
schroedinger? ( media-libs/schroedinger )
X? ( x11-libs/libX11 x11-libs/libXext )
amr? ( media-libs/amrnb media-libs/amrwb )"
|
You have to read the ebuild to determine that the mp3, vorbis, theora, x264 and xvid USE flags only have an effect if the 'encoding' USE flag is set. Then, you must read the ffmpeg documentation (or try it out) to know that decoding support for these formats is included by default. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
d2_racing Bodhisattva
Joined: 25 Apr 2005 Posts: 13047 Location: Ste-Foy,Canada
|
Posted: Wed Oct 29, 2008 11:40 am Post subject: |
|
|
That's bigger that Expat Hell for my concern. |
|
Back to top |
|
|
audiodef Watchman
Joined: 06 Jul 2005 Posts: 6640 Location: The soundosphere
|
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Wed Oct 29, 2008 1:19 pm Post subject: |
|
|
audiodef,
That's probably because devs only test package upgrades against latest versions of their deps. This is an even bigger issue with unstable arches. Hence the recommendation to do deep upgrades.
I prefer to use stable and try my luck with only leaf upgrades, along with whatever dep upgrades are pulled in. Fewer total upgrades, less breakage on a continuing basis; meta-package upgrades (e.g., GNOME) end up upgrading a lot of deps in either case, so I just wait for them.
***
Geralt/d2_racing,
Can you expect the devs to test every USE flag combination? The only reason this one hurt is because it broke something essential needed to emerge (wget), otherwise a revdep-rebuild would fix it. This is why Gentoo is harder to maintain than a binary distribution. The coordination problem was exacerbated by the kerberos dev being on vacation.
Two possible solutions: have system be self-contained and minimize USE flags (a la BSD), or make all tools needed to emerge static. The latter is tricky for Portage, which uses Python and wget. _________________ Personal overlay | Simple backup scheme |
|
Back to top |
|
|
natros n00b
Joined: 26 Sep 2003 Posts: 70
|
Posted: Wed Oct 29, 2008 1:38 pm Post subject: |
|
|
yngwin wrote: | viralex wrote: | I know that the label 'amd64' is right, but I'd like to see 'x86_64' instead! |
Not going to happen. Renaming an arch would be a painful process, and it's just not worth the trouble. Besides, we use x86 and i686 also for AMD Durons and Athlons, so why should there be a problem with amd64? |
Because users might think that AMD64 is for AMD only and IA64 is for Intel EMT64T. And users should come first. _________________ -=[abit ip35 pro core2quad q6600 (oc 3.0 ghz) 8 gb ram 2x320 raid0 ext4 xfx8800gt 512mb]=- |
|
Back to top |
|
|
Geralt Apprentice
Joined: 14 Jun 2008 Posts: 150
|
Posted: Wed Oct 29, 2008 6:57 pm Post subject: |
|
|
Hypnos wrote: | Geralt/d2_racing,
Can you expect the devs to test every USE flag combination? The only reason this one hurt is because it broke something essential needed to emerge (wget), otherwise a revdep-rebuild would fix it. This is why Gentoo is harder to maintain than a binary distribution. The coordination problem was exacerbated by the kerberos dev being on vacation.
|
That's not the point. If you look into the first bug report it was from august, long before this bug reached stable. So the developers knew that this problem exists (in unstable back then, but that's what unstable ist there for) or bette had to knew that there is such a bug. Letting a known bug which affects core-system libraries und utils (for gentoo wget is a core utility) just should not happen. It's very very bad publicity and shows a weak QA and at least for core components (in stable) there should be a very good QA. |
|
Back to top |
|
|
Hypnos Advocate
Joined: 18 Jul 2002 Posts: 2889 Location: Omnipresent
|
Posted: Wed Oct 29, 2008 7:24 pm Post subject: |
|
|
Geralt,
I agree that repository QA should have caught this unresolved dep for mit-krb5 and heimdal. The devs know this. The problem is that once a mistake is made, in this case pushing e2fsprogs{-lib} into stable before the requiring packages, there's no easy way to take it back because it's already been deployed in systems.
FWIW, mit-krb5-1.6.3-r4 has already gone stable; the total delay is ~40 hours.
Does this point to a systemic problem in the way stabilization is done? Maybe; I am not aware of a pattern. In any case, automated QA tools wouldn't hurt.
EDIT: Devs are instructed to use repoman before committing, as described here. Even with the "kerberos" USE flag set, repoman did not find this error (my Portage tree is synced between the e2fsprogs{-lib} and mit-krb5-1.6.3-r4 stabilizations). And on my Core2 machine it took 20+ minutes just to run -- impossible to test all USE flag combos. Finally, I do not have qualudis (the Paludis repo QA tool) installed. |
|
Back to top |
|
|
|