Re: [Mageia-dev] Question about backports: calibre (bug 1659)
On 13 June 2011 11:51, Radu-Cristian FOTESCU beranger...@yahoo.ca wrote: Therefore, I strongly believe that all calibre updates be packaged into updates, not backports, especially as there isn't any Mageia 2.0 as of yet. Once Mageia 2.0 is released, whatever newer calibre releases will be available might go into backports instead -- if at all. (Although I'd say that it should still be updated to updates for the whole supported time of 1.0.) What do you think? I think that before thinking of updating or backporting that new version, you could start by actually update the package in cauldron so that: 1) it can be tested... 2) we're sure we won't have packages in mga1/updates newer that those of mga2/release
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Radu-Cristian FOTESCU a écrit : André, No matter what my e-mail address is, I am not in Canada, but in Romania. Ok. So you're close (in time zone) to most contributors :) Anyway, I'll think of packaging once I fix some other issues. Right now I'm investigating a very peculiar crash in KCharSelect (an upstream issue), which actually means KCharSelect crashes when a bad font is used (DejaVu _is_ having some bad issues). As I am not familiar with Qt4/KDE development, it's a kinky issue. And the bug is not where it seems to be. (I can't report the bug right now, but if you want details, ask me.) I am stunned that such an application like KCharSelect can crash such badly and nobody fixes it (yes, to reproduce the bug you must know to identify the actual conditions, however there are some upstream bug reports about this crashes, poorly defined). This being said, CharMap in Windows _never_ crashed, in no version of Windows, whereas KCharSelect _always_ crashes, from KDE 4.0 onwards. If I won't be able to pinpoint the bug (yes, I want to fix it), I might reconsider one more time using KDE4 (hence Linux) on my laptop, as this is utterly ridiculous to have KCharSelect crashing ... Maybe KCharSelect wasn't updated for KDE 4 ? Why not use gucharmap ? It seems complete, is desktop-neutral, and has never given me any problems. It's on the regular mageia dvd's (but not the dual). And of course in the repositories. Problem solved :) When I started, I was able to package my favorite application to start with, hopefully you can do the same, if it's not too complicated. (Since you indicate that it doesn't have dependancies to/from other packages, I suspect that it would be relatively straight-forward.) It needs Python 2.7 and whatnot, but this is not an issue. (I've packaged some RPMs in 2009, just not for Mandriva, for EL5-compatible distros.) Python 2.7 is in Mageia 1. If you've already had a Mandriva packaging account, you already qualify to be a Mageia packager. Otherwise you have to be mentored, but being familiar with packaging, it should be a very quick process :) R-C -- André
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Radu-Cristian FOTESCU a écrit : Maybe KCharSelect wasn't updated for KDE 4 ? It was, and very much so. Before having a kcharselect binary in kdeutils, it had prepared especially for that purpose a kcharselect widget as part of kdeui, in kdelibs. Amazing contournement... Why not use gucharmap ? It seems complete, is desktop-neutral, and has never given me any problems. It's on the regular mageia dvd's (but not the dual). And of course in the repositories. Problem solved :) On the contrary, the idea is not to avoid fixing the bugs, but to face the issues, identify them, and pinpoint the root cause. OK. Good luck. Hope you filed a bug report. Don't forget that 10 people doing the same 10% of a solution is not as effective as 10 people cooperating :) André R-C -- André
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Radu-Cristian FOTESCU a écrit : From: andre999and...@laposte.net [...] ... Considering your concern for the application, maybe you would like to package it for Mageia. You could ensure that it is always up to date, and that it works properly, and is properly supported. (The packager is a key player in support.) Just because it is called a backport doesn't mean that it won't work. The packager mentoring program will help you get started :) -- André Well, first of all, I never liked the _concept_ of backports. Too many repositories, too complex tree already. One of the reasons I wasn't very fond of Mandriva (the other reason being the IaOra theme(s).) As Stormi suggested, you could consider backports as feature updates. (Whether or not the repository names change.) There is a certain logic for having separate backport repositories. It is normal to put more focus on security updates and bug fixes, than introducing new features. The former could also be considered release blockers, but never backports. So QA focuses on security updates and bug fixes. Also, Mandriva provided corporate support for the former, but not backports. Of course this concept doesn't apply to a volonteer community distro such as Mageia. Mageia policy is inherited from Mandriva, but is evidently subject to changes. In terms of support, the nature of support by Mageia is yet to be defined, but it is starting to be discussed. From the NON-rolling distros, Fedora is arguably the only one who tries to bring newer versions of a number of applications throughout its 12+1 months lifecycle. w/o using backports. My opinion is that, as long as system libraries are _not_ upgraded, many other packages (applications!) should be updated as appropriate. Otherwise, the result would be that Windows users would have more freedom and ease in decided what version of the [multi-platform open-source] applications to use than Linux users! (Except, of course, the users of rolling-release distros, and except for users of unstable/rawhide/cooker/cauldron...) I know, I should probably be using Fedora as long as _some_ of their principles suit my views much more than Mageia does or than Mandriva did. However, Fedora lacks something like Mandriva Control Center, and yum is millions of times slower than urpmi, therefore... I appreciate the same strengths inherited by Mageia. Not to mention that most of the best people Mandriva had are now with Mageia, which makes this distro hard to ignore... (Je crois qu'on appelle cela zugzwang...) I agree totally. Mageia is the best of the old Mandriva. So what I propose is that you seriously consider packaging your application for Mageia. We find a mentor for you to apprentice with, to familiarise you with the process. In choosing a mentor, it would help to find someone in the same time zone. You're in Canada ? What time zone ? (I'd offer to mentor you myself, being also in Canada, but I'm not yet a full packager.) When I started, I was able to package my favorite application to start with, hopefully you can do the same, if it's not too complicated. (Since you indicate that it doesn't have dependancies to/from other packages, I suspect that it would be relatively straight-forward.) Once you have started packaging, you have a better chance to influence Mageia policy, if you still think that it should be changed. But in any case you would be able to ensure that your package is available on Mageia, and is always up to date. And of course, ensure that it works properly. So, isn't it worth a try ? :) R-C -- André
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
So what I propose is that you seriously consider packaging your application for Mageia. We find a mentor for you to apprentice with, to familiarise you with the process. In choosing a mentor, it would help to find someone in the same time zone. You're in Canada ? What time zone ? (I'd offer to mentor you myself, being also in Canada, but I'm not yet a full packager.) André, No matter what my e-mail address is, I am not in Canada, but in Romania. Anyway, I'll think of packaging once I fix some other issues. Right now I'm investigating a very peculiar crash in KCharSelect (an upstream issue), which actually means KCharSelect crashes when a bad font is used (DejaVu _is_ having some bad issues). As I am not familiar with Qt4/KDE development, it's a kinky issue. And the bug is not where it seems to be. (I can't report the bug right now, but if you want details, ask me.) I am stunned that such an application like KCharSelect can crash such badly and nobody fixes it (yes, to reproduce the bug you must know to identify the actual conditions, however there are some upstream bug reports about this crashes, poorly defined). This being said, CharMap in Windows _never_ crashed, in no version of Windows, whereas KCharSelect _always_ crashes, from KDE 4.0 onwards. If I won't be able to pinpoint the bug (yes, I want to fix it), I might reconsider one more time using KDE4 (hence Linux) on my laptop, as this is utterly ridiculous to have KCharSelect crashing like shit (ask me and I'll tell you how to crash it on _any_ distro) and nobody doing anything! Millions of Linux users and developers! When I started, I was able to package my favorite application to start with, hopefully you can do the same, if it's not too complicated. (Since you indicate that it doesn't have dependancies to/from other packages, I suspect that it would be relatively straight-forward.) It needs Python 2.7 and whatnot, but this is not an issue. (I've packaged some RPMs in 2009, just not for Mandriva, for EL5-compatible distros.) R-C
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
So why does it have to be treated differently than the others since there is nothing special about this release cycle ? Michael, please give me an example of an application that releases on average 5 time a month. Really, give me an example. And, like it or not, calibre is _THE_ application for e-book lovers. For e-book users, it's as important as Firefox or Chromium is for the rest of the people. Also, Mageia 1.0 was released with a version of this application shamefully old. Fedora made a better judgment (I might just consider switching to Fedora, although I like Mageia more): they released F15 with a recent version of calibre, 0.7.56, but they added in updates 0.8.0. For the time being they stopped at 0.8.0, and only in Rawhide they pushed 0.8.4, which in my opinion is a good judgment that would be a _balance_ between: -- announced bug-fixes -- announced new features -- announced new hardware supported -- ad-hoc assessment of the risk brought by the new features (a heuristic process based mainly on experience as a user, experience as a software developer, and common-sense).Of course, the people who _make_ a distro are its _owners_, and of course the reasoning of those who make Fedora is not necessarily the best example to be followed by everybody, but as it happens, I prefer their brains. R-C
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Le mardi 14 juin 2011 à 03:00 -0700, Radu-Cristian FOTESCU a écrit : So why does it have to be treated differently than the others since there is nothing special about this release cycle ? Michael, please give me an example of an application that releases on average 5 time a month. Really, give me an example. Release frequency never was a criteria for differentiating between pushing something to updates and something to backports. And I see no reason why it would be in favor of doing a bug fix update rather than a backport, especially if we ask to do a more stringent QA checking on updates, as it would put too much work on the team. And, like it or not, calibre is _THE_ application for e-book lovers. For e-book users, it's as important as Firefox or Chromium is for the rest of the people. Again, that's not a criteria. Every software is important to at least one person, and that would mean we should update everything if we start to update everything important to one group of users. Also, Mageia 1.0 was released with a version of this application shamefully old. Fedora made a better judgment (I might just consider switching to Fedora, although I like Mageia more): they released F15 with a recent version of calibre, 0.7.56, but they added in updates 0.8.0. For the time being they stopped at 0.8.0, and only in Rawhide they pushed 0.8.4, which in my opinion is a good judgment that would be a _balance_ between: -- announced bug-fixes -- announced new features -- announced new hardware supported -- ad-hoc assessment of the risk brought by the new features (a heuristic process based mainly on experience as a user, experience as a software developer, and common-sense).Of course, the people who _make_ a distro are its _owners_, and of course the reasoning of those who make Fedora is not necessarily the best example to be followed by everybody, but as it happens, I prefer their brains. You still do not explain where is the problem of using backports for that. And for what it is worth, Fedora is discussing having separate update and backport ( https://fedorahosted.org/fesco/ticket/515 ), even if the discussion seems to be going nowhere at the moment -- Michael Scherer
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
martedì 14 giugno 2011 alle 12:00, Radu-Cristian FOTESCU ha scritto: Fedora made a better judgment I didn't find my commoncpp2 version there, for me is blocking, they ship 0.7.x and 0.8.x is out with *important* fixings! Every time a kernel update (upgrade) is pushed, we need to wait nvidia kernel modules or switch to free drivers... so all that glitters is not gold (well i hope that is the English meaning) Said that every distros have problems. Mageia has no monodevelop for me taht's blocking as well... Cheers, Angelo signature.asc Description: This is a digitally signed message part.
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
2011/6/14 Angelo Naselli anase...@linux.it martedì 14 giugno 2011 alle 12:00, Radu-Cristian FOTESCU ha scritto: . Said that every distros have problems. Mageia has no monodevelop for me taht's blocking as well... I will push monodevelop in svn asap (maybe this evening), for the revision of my mentor. Cheers, Angelo -- Ciao Stblack http://stblack.blogspot.com/
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Release frequency never was a criteria for differentiating between pushing something to updates and something to backports. It should be. Otherwise, we should all be using OpenOffice.org 1.0.1. -- security issues set aside. And I see no reason why it would be in favor of doing a bug fix update rather than a backport, especially if we ask to do a more stringent QA checking on updates, as it would put too much work on the team. Because Mageia (and Mandriva)'s vision of the concept of backports is not compatible with my common-sense. I have not used Mandriva very much in the past, because I hate the concept of backports -- yes, Ubuntu does them too, but Ubuntu backports are totally unsupported, so you can imagine their quality... I'd rather stick to updates -- this is also the reason I stopped using Debian, because the morons (yes, morons) were only pushing tzdata updates in volatile, not in updates, whereas ALL the other distro weres pushing tzdata updates in updates. If Mageia considers that a 6-7 months old package (for an application that released 32 times in the meantime) only deserves updates in backports, then I will probably stop reporting any possible bugs with this distro -- as a protest. It is indeed a matter of principle. I am personally using the latest calibre installed in /opt, not the official one, but again, it's a matter of principle. Whatever is important and comes from upstream should go into updates IMHO. Backports, in my view, only make sense if they're coming from Release N+1 *and* if they represent a major version bump -- such as FF4 over FF3.6, etc. WRT calibre, Fedora has a simple way: it keeps a newer calibre packages in updates/testing for 1 week, and if no user complains about regressions, it goes into updates. This is because calibre is a leaf package -- no other package depends on it, so it only impacts those who are using it. Again, that's not a criteria. Every software is important to at least one person, and that would mean we should update everything if we start to update everything important to one group of users. I can see how important is calibre to Mageia users. Nobody noticed or cared that it is an antiquated version. They could have as well used notepad.exe from Win95. And for what it is worth, Fedora is discussing having separate update and backport ( https://fedorahosted.org/fesco/ticket/515 ), even if the discussion seems to be going nowhere at the moment BS. I hope Fedora *never* uses backports! Their update policy is very clear *and* flexible: http://fedoraproject.org/wiki/Updates_Policy Please note these: Exceptions: Some classes of software will not fit in these guidelines. If your package does not fit in one of the classes below, but you think it should be allowed to update more rapidly . . . Things that would make it more likely to grant a request: -- The package is a leaf node. Nothing depends on it or requires it. Calibre is a leaf package. If not, in the same document: All other updates must either: -- reach the criteria laid out in the previous section OR -- reach the positive Bodhi karma threshold specified by the updates submitter OR -- spend some minimum amount of time in updates-testing, currently one week. I am not sure why F15 stopped updating calibre to 0.8.0 in updates (Rawhide went up to 0.8.4, maybe 0.8.5 now), but for the versions up to and including 0.8.0, here's the dynamics of the updates: ChangeLog: * Fri May 6 2011 Kevin Fenzi kevin@x - 0.8.0-1 - Update to 0.8.0 * Wed May 4 2011 Dan HorÃak dan@ - 0.7.59-2 - rebuilt against podofo 0.9.1 * Sat Apr 30 2011 Kevin Fenzi kevin@x - 0.7.59-1 - Update to 0.7.59 * Fri Apr 22 2011 Kevin Fenzi kevin@x - 0.7.57-1 - Update to 0.7.57 (F15 was released with 0.7.56) Indeed, Mageia does not have the number of packagers that Fedora has. However, if Mageia's _policy_ is to rather have 6-7 months old versions in updates, I should probably realize that Mageia is not for me. No, I have not, and never will use any repository called backports. When a newer stable release of a distro is available, I should update to it if updates I need are not pushed into Release N-1 updates (even if that release is officially still supported with security patches), but again, backports as Mandriva and Mageia are seeing them -- i.e. backporting from Cooker/Cauldron, not from updates/testing nor from Release N+1 -- does not fit my Zen. R-C
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
On 14 June 2011 15:52, Angelo Naselli anase...@linux.it wrote: martedì 14 giugno 2011 alle 12:00, Radu-Cristian FOTESCU ha scritto: Fedora made a better judgment I didn't find my commoncpp2 version there, for me is blocking, they ship 0.7.x and 0.8.x is out with *important* fixings! Every time a kernel update (upgrade) is pushed, we need to wait nvidia kernel modules or switch to free drivers... (FWIW I've not seen the nvidia proprietary kernel module fail to compile in about 6-8months, and the kernel got updated many times during the previous Cauldron release cycle). so all that glitters is not gold (well i hope that is the English meaning) Said that every distros have problems. Mageia has no monodevelop for me taht's blocking as well... Cheers, Angelo -- Ahmad Samir
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
On Tue, 14 Jun 2011, Radu-Cristian FOTESCU wrote: Indeed, Mageia does not have the number of packagers that Fedora has. However, if Mageia's _policy_ is to rather have 6-7 months old versions in updates, I should probably realize that Mageia is not for me. This will ultimately be up to the people who do the stable updates. As you can see from the fedora policy you quoted it is not possible to give exact rules for what kind of updates will be accepted. No, I have not, and never will use any repository called backports. When a newer stable release of a distro is available, I should update to it if updates I need are not pushed into Release N-1 updates (even if that release is officially still supported with security patches), but again, backports as Mandriva and Mageia are seeing them -- i.e. backporting from Cooker/Cauldron, not from updates/testing nor from Release N+1 -- does not fit my Zen. They will come from backports_testing I'd think. Likely packages will go from cauldron where they are tested a bit into backports_testing, and finally to backports. Would this 'backports' repository section sound better to you if it is renamed e.g. to 'rolling' ? (: Christiaan
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Le mardi 14 juin 2011 13:28:43, Ahmad Samir a écrit : On 14 June 2011 15:52, Angelo Naselli anase...@linux.it wrote: martedì 14 giugno 2011 alle 12:00, Radu-Cristian FOTESCU ha scritto: Fedora made a better judgment I didn't find my commoncpp2 version there, for me is blocking, they ship 0.7.x and 0.8.x is out with *important* fixings! Every time a kernel update (upgrade) is pushed, we need to wait nvidia kernel modules or switch to free drivers... (FWIW I've not seen the nvidia proprietary kernel module fail to compile in about 6-8months, and the kernel got updated many times during the previous Cauldron release cycle). Maybe he was talking about fglrx :) -- Balcaen John Jabber ID: mik...@jabber.littleboboy.net
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Would this 'backports' repository section sound better to you if it is renamed e.g. to 'rolling' ? (: No, backports sounds very bad, except maybe for people coming from Mandriva. They (you!) must love the concept. (Oh, maybe trolling instead of rolling :-)) Let me put it this way. People coming from Windows have this mind set: -- when I am using a release the OS called Windows XP, the system in itself gets minor updates: SP1, SP2, SP3. I therefore expect a release of a Linux distro to update KDE 4.6 to 4.6.1, 4.6.2, 4.6.3, 4.6.4... Now, KDE 4.8 would be more like upgrading XP to Vista so _no_, this should _not_ go into updates. -- when I am using Windows XP, applications that don't need newer libraries, like Calibre, can be updated as they are released. I don't need a new release of the OS to update such an application! (And no, it's _not_ backported from Vista or Win7!) So why is this impossible in Linux? (Of course, when Calibre will need a newer Python , it would be a different matter, but this is not the case yet.) Nwo, in Windows it's easy to downgrade in case of significant regressions. Everyone can do that. Not that easy in Linux though. If I remember correctly from the times when I was using Synaptic with Ubuntu, it was possible from a menu to choose, for each package, what version to install -- this way, downgrades were easy. Mandriva does not offer such a possibility. Either way, knowing that some major distros released with Linux _kernels_ that brought major regressions, and that _ALL_ the distros release every now and then X.Org with Intel/Nvidia/ATI drivers that _break_ (i.e. major regressions), I feel that so much fuss for a leaf application (that only breaks some features from itself, when this happens) is... too much fuss. R-C
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
I (FWIW I've not seen the nvidia proprietary kernel module fail to compile in about 6-8months, and the kernel got updated many times during the previous Cauldron release cycle). Maybe he was talking about fglrx :) and not in Cauldron but in fedora not the last kernel update though (but maybe we update it later :) ) -- Angelo signature.asc Description: This is a digitally signed message part.
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
On 13/06/11 10:51, Radu-Cristian FOTESCU wrote: Therefore, I strongly believe that all calibre updates be packaged into updates, not backports, especially as there isn't any Mageia 2.0 as of yet. The reference to backporting means backporting from Cauldron. Jim
Re: [Mageia-dev] Question about backports: calibre (bug 1659)
Two people asserted that a newer calibre package should go into backports, not updates. I said also that there could be exceptions... Therefore, I strongly believe that all calibre updates be packaged into updates, not backports, especially as there isn't any Mageia 2.0 as of yet. Once Mageia 2.0 is released, whatever newer calibre releases will be available might go into backports instead -- if at all. (Although I'd say that it should still be updated to updates for the whole supported time of 1.0.) Backports is meant from cauldron Angelo signature.asc Description: This is a digitally signed message part.