Re: KDE SC 4.8.4 important problems
On 06/13/2012 10:58 PM, Albert Astals Cid wrote: El Dimecres, 13 de juny de 2012, a les 17:43:10, Vishesh Handa va escriure: On Wed, Jun 13, 2012 at 2:34 PM, Sebastian Trügsebast...@trueg.de wrote: I am totally lost. What is the problem with that commit? AFAIK, the problem is that changes were made in both Soprano and KDElibs for 4.9, but for 4.8.4, they are using this changed kdelibs with the old soprano. Why did this happen? kdelibs is frozen so there should not be a commit that needs to be only in kdelibs 4.9 and not in kdelibs 4.8.4 Simple: it is a bugfix. But apparently that fix reveals another bug which I only fixed in Soprano 2.8. Cheers, Albert On 06/12/2012 07:40 PM, José Manuel Santamaría Lema wrote: Therefore, I tend to think the kdelibs commit mentioned in that bug report (**3d66d429cb48781881af3e735f2a80**af2114ab25) shouldn't have been done for 4.8.4. A confirmation from a Nepomuk developer would be highly appreciated. Just in case it isn't obvious enough, actually I meant this one shouldn't have been done for 4.8.4: http://quickgit.kde.org/index.**php?p=kdelibs.gita=commith=** a43824a200ed76cef48a90bd3aee85**47698b7c91http://quickgit.kde.org/index. php?p=kdelibs.gita=commith=a43824a200ed76cef48a90bd3aee8547698b7c91 which is the kdelibs one. I just pasted the wrong hash.
Re: KDE SC 4.8.4 important problems
On Wed, Jun 13, 2012 at 2:34 PM, Sebastian Trüg sebast...@trueg.de wrote: I am totally lost. What is the problem with that commit? AFAIK, the problem is that changes were made in both Soprano and KDElibs for 4.9, but for 4.8.4, they are using this changed kdelibs with the old soprano. On 06/12/2012 07:40 PM, José Manuel Santamaría Lema wrote: Therefore, I tend to think the kdelibs commit mentioned in that bug report (**3d66d429cb48781881af3e735f2a80**af2114ab25) shouldn't have been done for 4.8.4. A confirmation from a Nepomuk developer would be highly appreciated. Just in case it isn't obvious enough, actually I meant this one shouldn't have been done for 4.8.4: http://quickgit.kde.org/index.**php?p=kdelibs.gita=commith=** a43824a200ed76cef48a90bd3aee85**47698b7c91http://quickgit.kde.org/index.php?p=kdelibs.gita=commith=a43824a200ed76cef48a90bd3aee8547698b7c91 which is the kdelibs one. I just pasted the wrong hash. -- Vishesh Handa
Re: KDE SC 4.8.4 important problems
On Wed, Jun 13, 2012 at 2:34 PM, Sebastian Trüg sebast...@trueg.de wrote: I am totally lost. What is the problem with that commit? AFAIK, the problem is that changes were made in both Soprano and KDElibs for 4.9, but for 4.8.4, they are using this changed kdelibs with the old soprano. Not correct - at openSUSE, we are using Soprano 2.7.6 with kdelibs 4.8.4 (and virtuoso 4.1.5) and we get the crashes mentioned in Jose Maria's original mail. Particularly the gwenview crash is very easy to reproduce: #2 gwenview I can reproduce it just clicking an image and sometimes even just starting the application. Backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_gwenview_santa.kcras I don't know whether there is an issue with the kdelibs commit or whether the SC/BC issue Aaron fixed with TCPClient is relevant. I don't understand why the backtraces seemingly have nothing to do with Soprano either, unless (guess) a threading problem causes useless backtraces to be generated. Will
Re: KDE SC 4.8.4 important problems
El Dimecres, 13 de juny de 2012, a les 17:43:10, Vishesh Handa va escriure: On Wed, Jun 13, 2012 at 2:34 PM, Sebastian Trüg sebast...@trueg.de wrote: I am totally lost. What is the problem with that commit? AFAIK, the problem is that changes were made in both Soprano and KDElibs for 4.9, but for 4.8.4, they are using this changed kdelibs with the old soprano. Why did this happen? kdelibs is frozen so there should not be a commit that needs to be only in kdelibs 4.9 and not in kdelibs 4.8.4 Cheers, Albert On 06/12/2012 07:40 PM, José Manuel Santamaría Lema wrote: Therefore, I tend to think the kdelibs commit mentioned in that bug report (**3d66d429cb48781881af3e735f2a80**af2114ab25) shouldn't have been done for 4.8.4. A confirmation from a Nepomuk developer would be highly appreciated. Just in case it isn't obvious enough, actually I meant this one shouldn't have been done for 4.8.4: http://quickgit.kde.org/index.**php?p=kdelibs.gita=commith=** a43824a200ed76cef48a90bd3aee85**47698b7c91http://quickgit.kde.org/index. php?p=kdelibs.gita=commith=a43824a200ed76cef48a90bd3aee8547698b7c91 which is the kdelibs one. I just pasted the wrong hash.
Re: KDE SC 4.8.4 important problems
El Dimecres, 13 de juny de 2012, a les 14:47:08, Will Stephenson va escriure: On Wed, Jun 13, 2012 at 2:34 PM, Sebastian Trüg sebast...@trueg.de wrote: I am totally lost. What is the problem with that commit? AFAIK, the problem is that changes were made in both Soprano and KDElibs for 4.9, but for 4.8.4, they are using this changed kdelibs with the old soprano. Not correct - at openSUSE, we are using Soprano 2.7.6 with kdelibs 4.8.4 (and virtuoso 4.1.5) and we get the crashes mentioned in Jose Maria's original mail. Yes, that's the whole point, there is something wrong in the Soprano 2.7.6 + kdelibs 4.8.4 combination Soprano 2.7.6 + kdelibs 4.8.3 works Soprano 2.7.57 + kdelibs 4.8.4 works Cheers, Albert Particularly the gwenview crash is very easy to reproduce: #2 gwenview I can reproduce it just clicking an image and sometimes even just starting the application. Backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_gwenview_santa.kcr as I don't know whether there is an issue with the kdelibs commit or whether the SC/BC issue Aaron fixed with TCPClient is relevant. I don't understand why the backtraces seemingly have nothing to do with Soprano either, unless (guess) a threading problem causes useless backtraces to be generated. Will
Re: KDE SC 4.8.4 important problems
The relevant commits have been reverted in the KDE/4.8.x branch. This should fix the issues. On Thu, Jun 14, 2012 at 2:29 AM, Albert Astals Cid aa...@kde.org wrote: El Dimecres, 13 de juny de 2012, a les 14:47:08, Will Stephenson va escriure: On Wed, Jun 13, 2012 at 2:34 PM, Sebastian Trüg sebast...@trueg.de wrote: I am totally lost. What is the problem with that commit? AFAIK, the problem is that changes were made in both Soprano and KDElibs for 4.9, but for 4.8.4, they are using this changed kdelibs with the old soprano. Not correct - at openSUSE, we are using Soprano 2.7.6 with kdelibs 4.8.4 (and virtuoso 4.1.5) and we get the crashes mentioned in Jose Maria's original mail. Yes, that's the whole point, there is something wrong in the Soprano 2.7.6 + kdelibs 4.8.4 combination Soprano 2.7.6 + kdelibs 4.8.3 works Soprano 2.7.57 + kdelibs 4.8.4 works Cheers, Albert Particularly the gwenview crash is very easy to reproduce: #2 gwenview I can reproduce it just clicking an image and sometimes even just starting the application. Backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_gwenview_santa.kcr as I don't know whether there is an issue with the kdelibs commit or whether the SC/BC issue Aaron fixed with TCPClient is relevant. I don't understand why the backtraces seemingly have nothing to do with Soprano either, unless (guess) a threading problem causes useless backtraces to be generated. Will -- Vishesh Handa
Re: KDE SC 4.8.4 important problems
On 06/10/2012 12:57 PM, Andreas Pakulat wrote: Hi, Am Sonntag, 10. Juni 2012 schrieb Peter Penz : On 06/10/2012 11:20 AM, Aaron J. Seigo wrote: On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::__defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. The issue has been tracked at https://bugs.kde.org/show_bug.__cgi?id=268064 https://bugs.kde.org/show_bug.cgi?id=268064 - updating Soprano to the latest master resolves the crash. But I don't know more about the root-cause of this. Probably a Nepomuk-related update missed a proper versioning-check of Soprano? There has been an abi breakage in soparano's latest release (fixed in the repository already), so updating to that soprano release requires rebuilding all other code that uses it. I've seen backtraces ending in qstring::ref having such abi incompatibilities as cause, so it would fit at least those cases. Can someone explain how a missing class declaration (that is all that changed) can result in such a crash? I simply want to understand, not challenge it. Cheers, Sebastian
Re: KDE SC 4.8.4 important problems
I just wanted to note, that the root of all those crashes is not an ABI- breakage or a bug in kdelibs. It's once more a problem with threading and Qt Sockets. The reason why soprano-2.7.6 works is that it introduced Soprano::Socket. I already wrote it in another bug report, that I extracted a patch from the diff between 2.7.6 and 2.7.56, that simply introduces Soprano::Socket to the 2.7.6 Codebase - and all those crashes are gone, now running 2 days without a crash, without problems. Is it possible, to release soprano-2.7.6.1 or so, that ships with Soprano::Socket? This would make it possible for distributions to ship a working kde-4.8.4 with only stable software.
Re: KDE SC 4.8.4 important problems
Yeah. So Nepomuk is the cause of the problems - Here our our options - 1. I revert Sebastian's commits in kdelibs. This should fix the issue, but we would need to reintroduce the changes for 4.9, and since we do not have separate branches ... 2. Sebastian should release a new version (2.8) of Soprano any day now, packagers will need to get everyone to update. Or On Tue, Jun 12, 2012 at 5:57 PM, alpine.art...@googlemail.com wrote: I just wanted to note, that the root of all those crashes is not an ABI- breakage or a bug in kdelibs. It's once more a problem with threading and Qt Sockets. The reason why soprano-2.7.6 works is that it introduced Soprano::Socket. I already wrote it in another bug report, that I extracted a patch from the diff between 2.7.6 and 2.7.56, that simply introduces Soprano::Socket to the 2.7.6 Codebase - and all those crashes are gone, now running 2 days without a crash, without problems. We do as suggested over here. Is it possible, to release soprano-2.7.6.1 or so, that ships with Soprano::Socket? This would make it possible for distributions to ship a working kde-4.8.4 with only stable software. -- Vishesh Handa
Re: KDE SC 4.8.4 important problems
On 2012-06-12, Vishesh Handa m...@vhanda.in wrote: --bcaec554d60626569204c246cba9 Content-Type: text/plain; charset=ISO-8859-1 Yeah. So Nepomuk is the cause of the problems - Here our our options - 1. I revert Sebastian's commits in kdelibs. This should fix the issue, but we would need to reintroduce the changes for 4.9, and since we do not have separate branches ... 2. Sebastian should release a new version (2.8) of Soprano any day now, packagers will need to get everyone to update. 3. Actually track the bug down and fix it rather than try to do workarounds? This sounds like the most obvious thing to me. /Sune
Re: KDE SC 4.8.4 important problems
On Tue, Jun 12, 2012 at 7:58 PM, Sune Vuorela nos...@vuorela.dk wrote: On 2012-06-12, Vishesh Handa m...@vhanda.in wrote: --bcaec554d60626569204c246cba9 Content-Type: text/plain; charset=ISO-8859-1 Yeah. So Nepomuk is the cause of the problems - Here our our options - 1. I revert Sebastian's commits in kdelibs. This should fix the issue, but we would need to reintroduce the changes for 4.9, and since we do not have separate branches ... 2. Sebastian should release a new version (2.8) of Soprano any day now, packagers will need to get everyone to update. 3. Actually track the bug down and fix it rather than try to do workarounds? This sounds like the most obvious thing to me. I'm not familiar with the code or the changes that have been made, so I'm not going to try to do that right now. Probably. @Sebastian: Do you have the time? /Sune -- Vishesh Handa
Re: KDE SC 4.8.4 important problems
Therefore, I tend to think the kdelibs commit mentioned in that bug report (3d66d429cb48781881af3e735f2a80af2114ab25) shouldn't have been done for 4.8.4. A confirmation from a Nepomuk developer would be highly appreciated. Just in case it isn't obvious enough, actually I meant this one shouldn't have been done for 4.8.4: http://quickgit.kde.org/index.php?p=kdelibs.gita=commith=a43824a200ed76cef48a90bd3aee8547698b7c91 which is the kdelibs one. I just pasted the wrong hash. signature.asc Description: This is a digitally signed message part.
Re: KDE SC 4.8.4 important problems
On 06/12/2012 06:43 PM, Vishesh Handa wrote: On Tue, Jun 12, 2012 at 7:58 PM, Sune Vuorela nos...@vuorela.dk mailto:nos...@vuorela.dk wrote: On 2012-06-12, Vishesh Handa m...@vhanda.in mailto:m...@vhanda.in wrote: --bcaec554d60626569204c246cba9 Content-Type: text/plain; charset=ISO-8859-1 Yeah. So Nepomuk is the cause of the problems - Here our our options - 1. I revert Sebastian's commits in kdelibs. This should fix the issue, but we would need to reintroduce the changes for 4.9, and since we do not have separate branches ... 2. Sebastian should release a new version (2.8) of Soprano any day now, packagers will need to get everyone to update. 3. Actually track the bug down and fix it rather than try to do workarounds? This sounds like the most obvious thing to me. I'm not familiar with the code or the changes that have been made, so I'm not going to try to do that right now. Probably. @Sebastian: Do you have the time? I do not have much time. But the real problem is that I am not sure where to look. It has to do with my own implementation of unix socket communication. Someone with experience in that area might want to review the *Socket* classes in soprano/client... /Sune -- Vishesh Handa
Re: KDE SC 4.8.4 important problems
On 2012-06-12, Sebastian Trüg tr...@kde.org wrote: I do not have much time. But the real problem is that I am not sure where to look. It has to do with my own implementation of unix socket communication. Someone with experience in that area might want to review the *Socket* classes in soprano/client... This might be a stupid question, but isn't soprano 2.6 using QLocalSocket ? I was btw trying to track down what is going on. I'm still unsure why it is crashing, but a launch of e.g. gwenview is deleting recreating GlobalModelContainer::localSocketModel and related bits 6 times. in newer kdelibs (when init(true) is called) and I'm not sure if references to the old localSocketmodel can still be around. I agre with Jose Santamaria Lema that something in that commit looks a bit fishy. /Sune
Re: KDE SC 4.8.4 important problems
Hi, Peter Penz peter.pen...@gmail.com The issue has been tracked at https://bugs.kde.org/show_bug.cgi?id=268064 - updating Soprano to the latest master resolves the crash. But I don't know more about the root-cause of this. Probably a Nepomuk-related update missed a proper versioning-check of Soprano? There is also another bug report about all these crashes in general https://bugs.kde.org/show_bug.cgi?id=301453 Also, according to Manu Tortosa, indeed kde 4.8.4 works without crashes with the newer, beta soprano.. I inspected the soprano git repository and I found this commit: http://quickgit.kde.org/index.php?p=soprano.gita=commith=3d66d429cb48781881af3e735f2a80af2114ab25 which was done the same day than the commit mentioned in the above bug report https://bugs.kde.org/show_bug.cgi?id=301453#c2 So I built and install kdelibs reverting the commit mentioned in the comment #2, this hacked version of kdelibs has been making the crashes gone so far for me. I have checked the debug output for gwenview with a system with the official 4.8.4 kdelibs, I'm attaching it as gwenview_stdout_crashing.txt. I did the same with a system with the hacked kdelibs, I'm attaching that output as gwenview_stdout_working.txt Note that the working gwenview produces just one line like this: gwenview(3899)/nepomuk (library) {anonymous}::GlobalModelContainer::init: Connecting to local socket /tmp/ksocket-santa/nepomuk-socket while the crashing gwenview tries to connect to that socket various times. Also note that if I remove /tmp/ksocket-santa/nepomuk-socket the crashes are gone. So I also built a hacked soprano; this hacked soprano is 2.7.6 patched with these couple of commits: http://quickgit.kde.org/index.php?p=kdelibs.gita=commith=a43824a200ed76cef48a90bd3aee8547698b7c91 http://quickgit.kde.org/index.php?p=soprano.gita=commith=5dde64227186d405f86bf66256705fbceabcf1e2 Installing it in an affected system makes the crashes vanish. Therefore, I tend to think the kdelibs commit mentioned in that bug report (3d66d429cb48781881af3e735f2a80af2114ab25) shouldn't have been done for 4.8.4. A confirmation from a Nepomuk developer would be highly appreciated. $ gwenview gwenview(326)/nepomuk (library) {anonymous}::GlobalModelContainer::init: Connecting to local socket /tmp/ksocket-santa/nepomuk-socket gwenview(326)/nepomuk (library) {anonymous}::GlobalModelContainer::init: Connecting to local socket /tmp/ksocket-santa/nepomuk-socket gwenview(326)/nepomuk (library) {anonymous}::GlobalModelContainer::init: Connecting to local socket /tmp/ksocket-santa/nepomuk-socket gwenview(326)/kio (bookmarks) KBookmarkManager::KBookmarkManager: starting KDirWatch for /home/santa/.local/share/user-places.xbel gwenview(326)/kdecore (KSycoca) KSycocaPrivate::openDatabase: Trying to open ksycoca from /var/tmp/kdecache-santa/ksycoca4 gwenview(326)/kio (KDirListerCache) KDirListerCache::listDir: Listing directory: KUrl(trash:/) gwenview(326)/kio (Scheduler) KIO::SchedulerPrivate::doJob: KIO::SimpleJob(0x1b36470) gwenview(326)/kio (Scheduler) KIO::SchedulerPrivate::protoQ: creating ProtoQueue instance for trash gwenview(326)/kio (Scheduler) KIO::ProtoQueue::ProtoQueue: m_maxConnectionsTotal: 2 m_maxConnectionsPerHost: 2 gwenview(326)/kdecore (KLocale) KuitSemanticsPrivate::formatFromContextMarker: Unknown semantic role '@addaction' in context marker for message {Name}. gwenview(326)/kdecore (KLocale) KuitSemanticsPrivate::formatFromContextMarker: Unknown semantic role '@addaction' in context marker for message {Date}. gwenview(326)/kdecore (KLocale) KuitSemanticsPrivate::formatFromContextMarker: Unknown semantic role '@addaction' in context marker for message {Size}. gwenview(326)/nepomuk (library) {anonymous}::GlobalModelContainer::init: Connecting to local socket /tmp/ksocket-santa/nepomuk-socket gwenview(326)/kio (KDirListerCache) KDirListerCache::listDir: Entry currently being listed: KUrl(trash:/) by (KDirLister(0x1b36780) ) gwenview(326)/nepomuk (library) {anonymous}::GlobalModelContainer::init: Connecting to local socket /tmp/ksocket-santa/nepomuk-socket gwenview(326)/kdeui (kdelibs): Attempt to use QAction edit_redo with KXMLGUIFactory! gwenview(326)/kdeui (kdelibs): Attempt to use QAction edit_undo with KXMLGUIFactory! gwenview(326)/kdecore (K*TimeZone*) KSystemTimeZonesPrivate::instance: instance(): ... initialised gwenview(326)/kdecore (K*TimeZone*) KSystemTimeZonesPrivate::readConfig: readConfig(): local zone= Europe/Madrid gwenview(326)/kdecore (K*TimeZone*) KSystemTimeZonesPrivate::readZoneTab: readZoneTab( /usr/share/zoneinfo/zone.tab ) gwenview(326)/nepomuk (library) {anonymous}::GlobalModelContainer::init: Connecting to local socket /tmp/ksocket-santa/nepomuk-socket gwenview(326)/kio (KDirListerCache) KDirListerCache::listDir: Entry currently being listed: KUrl(trash:/) by (KDirLister(0x1b36780) , KDirLister(0x1de3c40) ) gwenview(326)/kio (KDirListerCache) KDirListerCache::stopListingUrl:
KDE SC 4.8.4 important problems
Hello, firstly, to be honest, let me say I'm doing huge effort to write this mail in a way that I don't look like a clueless cretin who reports bugs to the mailing lists instead of the proper place (bugs.kde.org) just to cheat and get my bugs fixed. I promise I'm not that kind of person, I have a git/svn KDE account and I'm already doing a very modest contribution. In my honest opinion I know a bit how things work in KDE with regarding to the bug reporting stuff. Also I'm packaging KDE for Debian in a team collaborating with other people since some time ago (I've been involved in 4.5.x, 4.6.x, 4.7.x and 4.8.x) and that's why I am writing here: when testing the latest point release of KDE SC (4.8.4), we are getting several crashes, and most of them are the kind of bugs (called release-critical in Debian) which would prevent us to release packages in the next stable release or migrate them to the testing branch. Most of these crashes follow this pattern (sorry for the vagueness, I hope to gather more information this night or so, I'll try to keep you posted but I can't promise anything): 1. Can't be reproduced with Nepomuk disabled. 2. Can't be reproduced with kdelibs 4.8.3 3. Can't be reproduced with KDE SC 4.9. 4. Could also be reproduced by various users and developers from other distributions. 5. They are not 100% reproducible; however they happen everytime in some computers, on others just very often; so often that make certain KDE programs unusuable. 6. I _feel_ backtraces might be invalid. Since I have been packaging KDE (more or less since 4.5 as said above) I never got 7 potential RC bugs in a release (not even a .0 one) after just ~2 days of use. Read again the characteristics 2. 3. and .4; that in conjunction with the fact that some people just discovered today how you are managing git branches/freeze for kdelibs (myself included) is why I think we are getting some people in other thread questioning -with or without reason- the kdelibs development model. I don't want to discuss the kdelibs development model in this thread (I would really prefer to keep any possible omfg kdelibs developers are using the same branch for 4.9 and 4.8 messages in the other thread) I'm just explaining the situation, I hope you get now why you are reading the things you are reading in the other thread. That being said, in this thread I want to focus on the actual grave problem(s) of 4.8.4, this is the list of crashes I have been experiencing: #1 dolphin: I can reproduce it right now in one of my computers just opening it, in other sometimes it crashes just clicking a file or folder. Eshat Cakar -a fellow Debian packager- also got other crash just clicking a folder, however he can't reproduce it anymore after he rebooted his computer. Backtraces: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_dolphin_santa.kcrash. http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_dolphin_eshat.txt #2 gwenview I can reproduce it just clicking an image and sometimes even just starting the application. Backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_gwenview_santa.kcrash #3 kontact executing the kaddressbook component I can reproduce it very oftenly just clicking the kaddressbok component in kontact. Backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_kontact_santa.kcrash #4 kmail executed outside kontact I can reproduce it just executing kmail outside kontact Backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug_kmail_santa.kcrash #5 kmail marking messages as read I think this one got fixed by a commit in kdepimlibs, which I included in the debian packaging. Please ignore for now. #6 kontact executing various components: calendar, to-do list, journal When clicking on various kontact component I'm getting a segmentation fault, always with the same backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/bug2_kontact_santa.kcrash #7 kmail links I can reproduce this one just opening a message with an url. Just clicking that url, which would it open it in konqueror given my local configuration crashes kmail: Backtrace: http://alioth.debian.org/~santa-guest/4.8.4_crashes/kmail_links.kcrash signature.asc Description: This is a digitally signed message part.
Re: KDE SC 4.8.4 important problems
On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. #3 kontact executing the kaddressbook component #4 kmail executed outside kontact these need to be sent to the kdepim team as they are crashes in kdepim components (address book library, kmail editor, ...) -- Aaron J. Seigo signature.asc Description: This is a digitally signed message part.
Re: KDE SC 4.8.4 important problems
On Sunday, June 10, 2012 11:20:09 Aaron J. Seigo wrote: On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. curious: on an affected system, if you delete `kde4-config --localprefix`/cache-freedom/ksycoca4* do the crashes go away? possible culprit commits: 1a07fea2eaa69d571d5818812502edcc1d680d9c e91e5fed6b1aad365e12e919f430c3e8147552d3 neither are super recent, but they both touch the relevant code. there was one change to kmimetype by dfaure but it is not implicated by the backtraces nor can i see how it would be (it added a required method, did not change existing code paths and definitely not the ksycoca code underneath it) -- Aaron J. Seigo signature.asc Description: This is a digitally signed message part.
Re: KDE SC 4.8.4 important problems
On 06/10/2012 11:20 AM, Aaron J. Seigo wrote: On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. The issue has been tracked at https://bugs.kde.org/show_bug.cgi?id=268064 - updating Soprano to the latest master resolves the crash. But I don't know more about the root-cause of this. Probably a Nepomuk-related update missed a proper versioning-check of Soprano? #3 kontact executing the kaddressbook component #4 kmail executed outside kontact these need to be sent to the kdepim team as they are crashes in kdepim components (address book library, kmail editor, ...)
Re: KDE SC 4.8.4 important problems
Den 10-06-2012 11:38, Peter Penz skrev: The issue has been tracked at https://bugs.kde.org/show_bug.cgi?id=268064 - updating Soprano to the latest master resolves the crash. But I don't know more about the root-cause of this. Probably a Nepomuk-related update missed a proper versioning-check of Soprano? I run a fully updated archlinux, and get these crashes too. So using kde 4.8.4 requires unreleased soprano ? :0 Anders
Re: KDE SC 4.8.4 important problems
Hi, Am Sonntag, 10. Juni 2012 schrieb Peter Penz : On 06/10/2012 11:20 AM, Aaron J. Seigo wrote: On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::**defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. The issue has been tracked at https://bugs.kde.org/show_bug.**cgi?id=268064https://bugs.kde.org/show_bug.cgi?id=268064- updating Soprano to the latest master resolves the crash. But I don't know more about the root-cause of this. Probably a Nepomuk-related update missed a proper versioning-check of Soprano? There has been an abi breakage in soparano's latest release (fixed in the repository already), so updating to that soprano release requires rebuilding all other code that uses it. I've seen backtraces ending in qstring::ref having such abi incompatibilities as cause, so it would fit at least those cases. Andreas
Re: KDE SC 4.8.4 important problems
Hello, On sekmadienis 10 Birželis 2012 13:57:09 Andreas Pakulat wrote: There has been an abi breakage in soparano's latest release (fixed in the repository already), so updating to that soprano release requires rebuilding all other code that uses it. I've seen backtraces ending in qstring::ref having such abi incompatibilities as cause, so it would fit at least those cases. We have never had soprano 2.5.76 or other alpha, only stable 2.7.6. And kdelibs 4.8.4 is completely broken for us. -- Modestas Vainius mo...@debian.org signature.asc Description: This is a digitally signed message part.
Re: KDE SC 4.8.4 important problems
El Diumenge, 10 de juny de 2012, a les 11:30:13, Aaron J. Seigo va escriure: On Sunday, June 10, 2012 11:20:09 Aaron J. Seigo wrote: On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. curious: on an affected system, if you delete `kde4-config --localprefix`/cache-freedom/ksycoca4* do the crashes go away? No they don't Cheers, Albert possible culprit commits: 1a07fea2eaa69d571d5818812502edcc1d680d9c e91e5fed6b1aad365e12e919f430c3e8147552d3 neither are super recent, but they both touch the relevant code. there was one change to kmimetype by dfaure but it is not implicated by the backtraces nor can i see how it would be (it added a required method, did not change existing code paths and definitely not the ksycoca code underneath it)
Re: KDE SC 4.8.4 important problems
El Diumenge, 10 de juny de 2012, a les 12:57:09, Andreas Pakulat va escriure: Hi, Am Sonntag, 10. Juni 2012 schrieb Peter Penz : On 06/10/2012 11:20 AM, Aaron J. Seigo wrote: On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::**defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. The issue has been tracked at https://bugs.kde.org/show_bug.**cgi?id=268064https://bugs.kde.org/show_bu g.cgi?id=268064- updating Soprano to the latest master resolves the crash. But I don't know more about the root-cause of this. Probably a Nepomuk-related update missed a proper versioning-check of Soprano? There has been an abi breakage in soparano's latest release (fixed in the repository already), so updating to that soprano release requires rebuilding all other code that uses it. I've seen backtraces ending in qstring::ref having such abi incompatibilities as cause, so it would fit at least those cases. As Modestias says, this has nothing to do with those ABI changes since it fails with stable soprano. Albert Andreas
Re: KDE SC 4.8.4 important problems
El Diumenge, 10 de juny de 2012, a les 11:49:38, Anders Lund va escriure: Den 10-06-2012 11:38, Peter Penz skrev: The issue has been tracked at https://bugs.kde.org/show_bug.cgi?id=268064 - updating Soprano to the latest master resolves the crash. But I don't know more about the root-cause of this. Probably a Nepomuk-related update missed a proper versioning-check of Soprano? I run a fully updated archlinux, and get these crashes too. So using kde 4.8.4 requires unreleased soprano ? :0 It should not, but seems something bad slipped in. Cheers, Albert Anders
Re: KDE SC 4.8.4 important problems
El Diumenge, 10 de juny de 2012, a les 11:20:09, Aaron J. Seigo va escriure: On Sunday, June 10, 2012 03:23:04 José Manuel Santamaría Lema wrote: #1 dolphin: #2 gwenview #6 kontact executing various components: calendar, to-do list, journal #7 kmail links these are all the same crash, or at least related to each other. it is crashing in KServiceTypeTrader::defaultOffers or KMimeTypeTrader::query apparently at times in KSycocaDict::find_string. FWIW if i add Q_ASSERT(!m_factories.hasLocalData()); to KSycocaFactorySingleton constructor in ksycocafactory.h if does assert, don't ask me why since it makes no sense, and it doesn't make any sense either that it does not assert if nepomukserver is not running. I tried to do some debugging yesterday and the only conclusion i found is that something is writing very off memory, but totally failed which could be the culprit and gdb/valgrind did not point me in the right direction either (or i cculd not read their hints if there was any :D) Cheers, Albert #3 kontact executing the kaddressbook component #4 kmail executed outside kontact these need to be sent to the kdepim team as they are crashes in kdepim components (address book library, kmail editor, ...)