| View previous topic :: View next topic |
| Author |
Message |
Nsane457
n00b
Joined: 25 Nov 2004
Posts: 45
|
 Posted: Tue May 10, 2016 3:24 am Post subject: Will GCC-6.1 be available anytime soon? |
 |
|
GCC-6.1.0 was released April 27th, 2016. I noticed that there was no ebuild, even a hard-masked ebuild, available in either gentoo or toolchain overlay for testing. I added it to my local overlay myself.
Is there a plan to releases an ebuild soon?
Also, I've been submitting patches for fixing gcc-6 build issues for various packages as "GCC Porting" bug reports. Are these even useful or desired? |
|
| Back to top |
|
 |
ct85711
Veteran
Joined: 27 Sep 2005
Posts: 1791
|
| Posted: Tue May 10, 2016 3:32 am Post subject: |
|
|
| well doing a quick search in the bugzilla, comes out with a tracker for gcc 6, so I would assume making bug report(s) for the applicable packages block the tracker with patches attached would be helpful. |
|
| Back to top |
|
|
Nsane457
n00b
Joined: 25 Nov 2004
Posts: 45
|
| Posted: Tue May 10, 2016 4:45 am Post subject: |
|
|
| ct85711 wrote: | | well doing a quick search in the bugzilla, comes out with a tracker for gcc 6, so I would assume making bug report(s) for the applicable packages block the tracker with patches attached would be helpful. |
Thanks for the link. I'll adjust the bug reports accordingly. |
|
| Back to top |
|
|
viralex
Apprentice
Joined: 24 Apr 2008
Posts: 237
Location: Viareggio (Lu,Italy)
|
| Posted: Tue May 10, 2016 2:28 pm Post subject: |
|
|
I'm testing gcc-6.1 on my personal repository on github ( it's not in a good shape but here it is : https://github.com/viralex/viralex-overlay )
I've just copyed 5.3.0 ebuild, renamed to 6.1.0 removed some lines.. done!
it works. if something goes horribly wrong tty or chroot  |
|
| Back to top |
|
|
derk
Guru
Joined: 10 Mar 2003
Posts: 347
Location: St Thomas Ontario
|
| Posted: Tue May 10, 2016 7:57 pm Post subject: |
|
|
| check-out hardened-development overlay or musl overlay for other ebuilds for gcc-6.1.0 |
|
| Back to top |
|
|
tranquilcool
Veteran
Joined: 25 Mar 2005
Posts: 1179
|
| Posted: Wed May 11, 2016 11:58 am Post subject: |
|
|
| derk wrote: | | check-out hardened-development overlay or musl overlay for other ebuilds for gcc-6.1.0 |
compiles fine, but glibc fails to build.
_________________
this is a strange strange world. |
|
| Back to top |
|
|
zorry
Developer
Joined: 30 Mar 2008
Posts: 380
Location: Umeå The north part of scandinavia
|
|
| Back to top |
|
|
lutel
Tux's lil' helper
Joined: 19 Oct 2003
Posts: 110
Location: Pomroczna
|
| Posted: Wed Sep 07, 2016 3:02 pm Post subject: |
|
|
| Why there is still no gcc-6, even masked in official portage? |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Sat Sep 10, 2016 8:58 pm Post subject: |
|
|
| lutel wrote: | | Why there is still no gcc-6, even masked in official portage? |
From the tracker bug above: Firefox, Chromium, gentoo-sources, glibc, Qt4, wxGTK. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6747
|
| Posted: Sat Sep 10, 2016 11:32 pm Post subject: |
|
|
| Ant P. wrote: | | lutel wrote: | | Why there is still no gcc-6, even masked in official portage? |
From the tracker bug above: Firefox, Chromium, gentoo-sources, glibc, Qt4, wxGTK. |
For all of these - practically for everything in the tracker bug and in all packages I compiled since then - there are trivial fixes (by adding some CFLAGS), for most of them even clean fixes by patches.
Moreover, even if there would be no patches/fixes, adding a masked gcc-6.2.0 would be a non-issue: With previous gcc versions there were sometimes really serious issues, but nevertheless the gcc versions were (unkeyworded or masked) in the gentoo repository a few days after they had appeared. |
|
| Back to top |
|
|
lutel
Tux's lil' helper
Joined: 19 Oct 2003
Posts: 110
Location: Pomroczna
|
| Posted: Sun Sep 11, 2016 12:44 pm Post subject: |
|
|
| mv wrote: | | Ant P. wrote: | | lutel wrote: | | Why there is still no gcc-6, even masked in official portage? |
From the tracker bug above: Firefox, Chromium, gentoo-sources, glibc, Qt4, wxGTK. |
For all of these - practically for everything in the tracker bug and in all packages I compiled since then - there are trivial fixes (by adding some CFLAGS), for most of them even clean fixes by patches.
Moreover, even if there would be no patches/fixes, adding a masked gcc-6.2.0 would be a non-issue: With previous gcc versions there were sometimes really serious issues, but nevertheless the gcc versions were (unkeyworded or masked) in the gentoo repository a few days after they had appeared. |
Its been very long time since GCC 6 was published, this situation is odd, especially for distribution where compiler is its heart. How compiler issues can be solved, if package maintainer doesn't even allow for masked ebuild to appear and encourage people to experiment with it? I'm starting to think it is done deliberately to hinder Gentoo development. Gentoo needs some management mechanism to prevent such hindrances, especially in such crucial parts as compiler. |
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
| Posted: Sun Sep 11, 2016 12:55 pm Post subject: |
|
|
| lutel wrote: | | Its been very long time since GCC 6 was published, this situation is odd, especially for distribution where compiler is its heart. |
...or maybe _because_ the compiler is its heart, terrible things will happen if it is unleashed too early. GCC-5, as it is now, is still not ready for stabilisation because of work needed on reverse dependencies. Anyone wanting to try out GCC-6 can do so already, but must be willing to find fixes to bugs and report them upstream as well. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6747
|
| Posted: Sun Sep 11, 2016 1:29 pm Post subject: |
|
|
| asturm wrote: | | ...or maybe _because_ the compiler is its heart, terrible things will happen if it is unleashed too early. |
Nobody was talking about unmasking or even keywording unstable.
| Quote: | | Anyone wanting to try out GCC-6 can do so already |
Not really, since it is not even clear which version will become (roughly) the one going into the gentoo tree. For instance, in previous releases the version from the hardened overlay often had a completely different patch set than the version in the main tree.
It is really a serious problem that development in gentoo has shifted completely to non-official overlays. This can make sense for huge projects like KDE or Gnome with dozens or hundreds of ebuilds for which masking or unmasking alone can be a major job which developers may want to avoid at the early testing stage. But for a single ebuild like gcc, it is really not understandable and limits the number of early testers (who would otherwise file bugs) for absolutely no reason. Indeed, it seems that it is an intention to limit this number of testers. |
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
| Posted: Sun Sep 11, 2016 2:02 pm Post subject: |
|
|
| Well, just trying to explain why it is not a given for Gentoo to be up front with new GCC versions, compared to binary distributions where the users do not actually depend on it for most of their package base. I don't know the reason for what's holding back GCC-6, as I'm not terribly obsessed with getting the latest version there - quite the contrary, I've been building with 4.8.5 until recently to detect packages that already require newer versions. |
|
| Back to top |
|
|
Buffoon
Veteran
Joined: 17 Jun 2015
Posts: 1369
Location: EU or US
|
| Posted: Sun Sep 11, 2016 2:27 pm Post subject: |
|
|
| Slackware is up to 6.2. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6747
|
| Posted: Sun Sep 11, 2016 2:36 pm Post subject: |
|
|
| asturm wrote: | | as I'm not terribly obsessed with getting the latest version there |
You would perhaps think differently if you would have C or C++ projects - it is almost mandatory to test them with the latest compiler versions to find hidden issues.
| Quote: | | I've been building with 4.8.5 until recently |
I had no issues with gcc-5 since more than a year. In fact, the change to gcc-5 as well as the recent change to gcc-6 were much less problematic than some gcc-4.x upgrades. You might be right, however, that you must still use unstable versions of one or two packages if you want to compile them with gcc-5. Some packages just lag for several years of development.
This sounds worse as it is: In my systems, only one single package had to be marked unstable: dosemu; its stable version is 7 years behind.
Well, lagging 7 years behind actually is a big issue, and this points to another problem of gentoo: Stabilization teams simply don't catch up, making it really dangerous to run a stable gentoo system. For instance, it is rather unlikely that such a 7 years old package has no security issues; just nobody had ever looked at the code, especially since upstream just had proceeded with development since then. Being even more conservative like Debian stable makes only sense if you do permanent security auditing which is actually not happening. The reason why such long outstanding stabilizations do not happen is simply lack of manpower. |
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
| Posted: Sun Sep 11, 2016 2:49 pm Post subject: |
|
|
| mv wrote: | | In fact, the change to gcc-5 as well as the recent change to gcc-6 were much less problematic than some gcc-4.x upgrades. |
If you know what you do, sure. GCC-5.4.0 works really well. But to this date, despite the news item, wiki and countless forum entries, you get all the same support questions over and over again because of the ABI breakage.
| mv wrote: | | You might be right, however, that you must still use unstable versions of one or two packages if you want to compile them with gcc-5. Some packages just lag for several years of development. |
I've actually got a fix pending for months already to get an ~arch package to build with GCC-5.4.0. And I know of several stable packages not ready for GCC-5, some where stabilisation was simply not possible until recently, and some that yeah, suffer from lack of stabilisation manpower. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6747
|
| Posted: Sun Sep 11, 2016 3:06 pm Post subject: |
|
|
| asturm wrote: | | because of the ABI breakage |
This has nothing to do with stabilization. In fact, postponing stabilization forever couldn't fix this problem. One simply must recompile the whole system after the gcc bump. Yes, that people forgot this simple fact which was well-known since almost the beginning of gentoo and which was in the (old?) gentoo Wiki for many years, is a problem. Even more that people just ignore news messages.
But again: This is a different problem and is not solved by postponing stabilization or by keeping masked packages in overlays outside of the tree.
| Quote: | | suffer from lack of stabilisation manpower |
Stabilization used to happen about 1-2 months after the package appeared in the tree. Meanwhile, less than a year is already an exception, 3-4 years is probably the average; some get never stabilized or even unkeyworded despite corresponding bug reports for years. (E.g. dev-tex/biber is still not keyworded ~x86 although it is a plain perl package - hence very unlikely not to work when it works on ~amd64 - and a corresponding bug report is there at least since months). Not to mentioned that many packages simply appear to never get stabilized. This is really becoming a problem. It makes stable systems less and less usable if you must have some hundreds of entries in accept_keywords, only for being able to run a reasonable desktop. |
|
| Back to top |
|
|
ct85711
Veteran
Joined: 27 Sep 2005
Posts: 1791
|
| Posted: Sun Sep 11, 2016 3:16 pm Post subject: |
|
|
| Honestly, I don't think it's necessarily a thing of lack of stabilization manpower which is the problem, but more of no one really bothering to do anything in general. I've been noticing that there is several bugs reports that have fixes to them (some where even up stream has accepted even); yet no one is willing to do do anything. Then there's the issue of where we are pushing more people and stuff to unofficial overlays. Since when did we stop supporting the Official tree, or is the devs too lazy to work on the official one and have to use the unofficial ones. |
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
| Posted: Sun Sep 11, 2016 3:25 pm Post subject: |
|
|
| ct85711 wrote: | | Honestly, I don't think it's necessarily a thing of lack of stabilization manpower which is the problem |
If you're involved in stabilisation bugs, then you see that it is exactly the problem. And if you have stabilisation struggling, you will see less stabilisation bugs filed by devs. |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Sun Sep 11, 2016 3:41 pm Post subject: |
|
|
| Personally, I would prefer the system being a year behind Debian, over some of the completely un-QAed updates williamh and vapier have been inflicting on Gentoo as of late. |
|
| Back to top |
|
|
mv
Watchman
Joined: 20 Apr 2005
Posts: 6747
|
| Posted: Sun Sep 11, 2016 5:21 pm Post subject: |
|
|
| Ant P. wrote: | | Personally, I would prefer the system being a year behind Debian |
I need my system for productive work and not for philosophical debates. And IMHO, the lag of gentoo is meanwhile reaching a critical stage: I am afraid the distribution is close to its death.
Just to give one example: When you have a problem with TeX, all tex users will tell you to use the current texlive-2016 - after all, it is out since almost 3/4 year, and already a few days later there was a Debian package. In Gentoo - although it is already only using the Debian tarballs for texlive - there is no sign that it will become available: No overlay exists, the bugs are without reply since the beginning. My guess is that the TeX team of gentoo is already dead.
I already have the feeling that I maintain half of the distribution in my overlay; I cannot take even more packages there and do the necessary bumps. But what to do when I need an at least semi-current texlive-2016? In the moment, when I have to install another distribution just to keep up with my daily work, I will not return to gentoo - there is no point in it, then.
This gcc-6 issue is now not the latest thing which shows that gentoo has a serious problem: Also glibc and binutils are lagging.
And why is an important package like palemoon still only in some overlay?
| Quote: | | completely un-QAed updates williamh and vapier have been inflicting |
Release fast, release often. Sure, in testing bugs can happen - that's what testing is for. Yes, it is a problem that vapier e.g. used to push completely unaudited eclass changes (or pushed them despite protests). But at least they used to do something which is - whether you like it or not - very important for a distribution to stay alive. Actually, it seems vapier also has quit, meanwhile. Everything seens to have stalled. Sure, there's summer holiday in some countries, but the stalling period is already much longer than just holidays: As mentioned, for texlive e.g. already 3/4 year or so.... |
|
| Back to top |
|
|
ct85711
Veteran
Joined: 27 Sep 2005
Posts: 1791
|
| Posted: Sun Sep 11, 2016 6:04 pm Post subject: |
|
|
| asturm wrote: | | If you're involved in stabilisation bugs, then you see that it is exactly the problem. And if you have stabilisation struggling, you will see less stabilisation bugs filed by devs. |
I'm not looking at just the stable branch, but also unstable and masked too... Like mv just mentioned, there's quite a few packages that haven't even been updated in the tree for new versions.
Sure stabilization is an issue, but it's only part of the overall problem. Take like the new plasma/kde... according the the forums, the current version in the tree has significant issues for multi-displays and other stuff; and the only solution is use some unofficial overlay to get a patch or updated ebuilds to maybe fix it...
https://bugs.gentoo.org/reports.cgi?product_id=&datasets=UNCONFIRMED&datasets=CONFIRMED&datasets=IN_PROGRESS&datasets=RESOLVED&datasets=FIXED
Take like this chart (Chart from Gentoo's buzgilla over time, showing the number of bugs reported, and fixed/closed). As the chart indicates, for a while Gentoo has been doing pretty good on resolving issues, but since roughly about 2015, the number of bugs that has been resolved/fixed has almost flat lined.
https://bugs.gentoo.org/reports.cgi?product_id=&datasets=UNCONFIRMED&datasets=CONFIRMED&datasets=IN_PROGRESS
Even here, (similar graph but only the opened bugs); the number of bugs that are open is also nearly flat lines since about the beginning of 2015.
Note: You can access these reports and similar, but going to Gentoo's Bugzilla, reports, and go from there... |
|
| Back to top |
|
|
asturm
Developer
Joined: 05 Apr 2007
Posts: 8935
|
| Posted: Sun Sep 11, 2016 6:16 pm Post subject: |
|
|
| ct85711 wrote: | | Sure stabilization is an issue, but it's only part of the overall problem. Take like the new plasma/kde... according the the forums, the current version in the tree has significant issues for multi-displays and other stuff; and the only solution is use some unofficial overlay to get a patch or updated ebuilds to maybe fix it... |
That's not true, KDE coverage in Gentoo is among the best of all distributions. Both latest releases of Plasma and Frameworks are in tree, okay yesterday's Frameworks release will probably make it tomorrow, and KDE Applications 16.08.1 is going to be moved soon. On top of that, Plasma-4 still being provided, and you don't want to know how messy it is to have both present... Multi-display issues is the same for everyone.
| mv wrote: | | I already have the feeling that I maintain half of the distribution in my overlay; |
If you do that already, why not push the work upstream? It is easier than ever to contribute via Gentoo's github mirror. That's how it started for me... |
|
| Back to top |
|
|
lutel
Tux's lil' helper
Joined: 19 Oct 2003
Posts: 110
Location: Pomroczna
|
| Posted: Sun Sep 11, 2016 6:59 pm Post subject: |
|
|
| mv wrote: |
I need my system for productive work and not for philosophical debates. And IMHO, the lag of gentoo is meanwhile reaching a critical stage: I am afraid the distribution is close to its death.
This gcc-6 issue is now not the latest thing which shows that gentoo has a serious problem: Also glibc and binutils are lagging.
And why is an important package like palemoon still only in some overlay?
Release fast, release often. Sure, in testing bugs can happen - that's what testing is for. Yes, it is a problem that vapier e.g. used to push completely unaudited eclass changes (or pushed them despite protests). But at least they used to do something which is - whether you like it or not - very important for a distribution to stay alive. Actually, it seems vapier also has quit, meanwhile. Everything seens to have stalled. Sure, there's summer holiday in some countries, but the stalling period is already much longer than just holidays: As mentioned, for texlive e.g. already 3/4 year or so.... |
I fully support this statement, I couldn't phrase it better. It is unbelievable that such distribution can die just because package maintainers can't cope with releases. Main strength of Gentoo is its ability to fine tune performance, security and stay on the edge. Maybe this is a call to rethink Gentoo management, if it must stay decendralised, there should be possibility to take away maintainer rights from people who no longer care about it, have open positions for such maintainers and keep a track on this process so all packages would have active maintainers. This is crucial for this distribution. Lot of scientific, academic and research entities run on Gentoo, but nobody seems to care about its quality. If you think Gentoo is slowly facing it lagging death, maybe some Gentoo Gods should initiate debate and start looking for financing or business owner who would care about its future. As for now, there should be taken some serious actions with GCC maintainers, as to me it looks we may have someone who is deliberately putting stick in Gentoo wheel, this issue of having masked GCC-6 build has been raised too many times. Someone just doesn't want Gentoo to develop, I can't see any other reasonable motive for that. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Unsupported Software
