Bug#429533: [pkg-boost-devel] Bug#429533: Bug#429533: Bug#429533: dev package changes and its severity
[Roger Leigh] I have brought this up with upstream. Please see the thread here: http://thread.gmane.org/gmane.comp.lib.boost.user/28950 and the bug report here: http://svn.boost.org/trac/boost/ticket/1094 Apparently, the library naming scheme on Linux should be layout=system, which uses the simple and expected naming scheme. I don't believe that is the concensus view. I think Neal Becker got it right [http://article.gmane.org/gmane.comp.lib.boost.devel/162674] I think typically linux distros want the version, but not the 'gcc41' part. We really do need to allow multiple versions, as is common practice with other shared libs on linux systems. Since multiple compilers are uncommon, distros normally ignore that. The trouble with --layout=system is that it completely strips the boost version from the library SONAME. Doing that will cause much more grief after an upgrade (because Boost doesn't maintain ABI) than the current crisis. But what is the simple and expected naming scheme to which you refer? Built using --layout=system, I find libboost_signals.so and libboost_signals-mt.so whereas the current Debian install has libboost_signals-st.so and libboost_signals-mt.so Notwithstanding the extra -st (which I don't think is a good idea), I think Debian's scheme is just fine. Is it the extra -st that is bugging you? Thanks, -Steve signature.asc Description: Digital signature
Bug#429533: [pkg-boost-devel] Bug#429533: Bug#429533: dev package changes and its severity
On Wed, Jun 27, 2007 at 05:30:21AM +0900, Junichi Uekawa wrote: Hi, Please elaborate 'right library name' Boost library naming is described in the documentation [0]. [0] http://www.boost.org/more/getting_started/unix-variants.html#library-naming Which does not mean it is correct for libboost-*-dev to silently change its interface (.so library naming). This description sounds more relevant for cross-platform development and other things, but isn't it possible to provide a saner default for gcc/libc system.s Why not just provide -pthread version (-mt) per default and provide -st version for people who really care? I have brought this up with upstream. Please see the thread here: http://thread.gmane.org/gmane.comp.lib.boost.user/28950 and the bug report here: http://svn.boost.org/trac/boost/ticket/1094 Apparently, the library naming scheme on Linux should be layout=system, which uses the simple and expected naming scheme. I haven't yet seen this documented, however. This solves GNU/Linux software library naming for linking, but does not fix it for true cross-platform use. The above thread and bug report propose a solution for that as well. Regards, Roger -- .''`. Roger Leigh : :' : Debian GNU/Linux http://people.debian.org/~rleigh/ `. `' Printing on GNU/Linux? http://gutenprint.sourceforge.net/ `-GPG Public Key: 0x25BFB848 Please GPG sign your mail. signature.asc Description: Digital signature
Bug#429533: [pkg-boost-devel] Bug#429533: Bug#429533: dev package changes and its severity
Hi, I have brought this up with upstream. Please see the thread here: http://thread.gmane.org/gmane.comp.lib.boost.user/28950 and the bug report here: http://svn.boost.org/trac/boost/ticket/1094 Apparently, the library naming scheme on Linux should be layout=system, which uses the simple and expected naming scheme. I haven't yet seen this documented, however. This solves GNU/Linux software library naming for linking, but does not fix it for true cross-platform use. The above thread and bug report propose a solution for that as well. What is said in the upstream sounds completely sound, and what has been said from the Debian maintainer sounds completely bogus. Ermm... so, what's the problem? Why not fix it? regards, junichi -- [EMAIL PROTECTED],netfort.gr.jp} Debian Project -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: dev package changes and its severity
Hello, Boost library naming is described in the documentation [0]. [0] http://www.boost.org/more/getting_started/unix-variants.html#library-naming Which does not mean it is correct for libboost-*-dev to silently change its interface (.so library naming). Moreover, if you have carefully read the document you're pointing to, you'd find there's nothing about -st. Only -mt or absence of -mt Regards Jiri Palecek -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: [pkg-boost-devel] Bug#429533: dev package changes and its severity
On Thu, Jul 19, 2007 at 12:08:08AM +0200, Ji??í Pale??ek wrote: Hello, hi, Boost library naming is described in the documentation [0]. [0] http://www.boost.org/more/getting_started/unix-variants.html#library-naming Which does not mean it is correct for libboost-*-dev to silently change its interface (.so library naming). Moreover, if you have carefully read the document you're pointing to, you'd find there's nothing about -st. Only -mt or absence of -mt i am aware of this and i don't like -st that much. before having -st, there was only absence of any -st and -mt which pointed to -mt. quite disgusting. cheers, domenico -[ Domenico Andreoli, aka cavok --[ http://www.dandreoli.com/gpgkey.asc ---[ 3A0F 2F80 F79C 678A 8936 4FEE 0677 9033 A20E BC50 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: [pkg-boost-devel] Bug#429533: dev package changes and its severity
On Tue, Jun 19, 2007 at 11:17:02PM +0900, Junichi Uekawa wrote: Hi, hi, I've browsed through the bug logs, but this is ugly. yes, it is We're breaking all those APIs, and there is no sane way of finding out what to use. we? who? breaking those APIs? it is only a matter of fixing the already broken library name used to link with programs. i don't see anything broken. Also, please keep the severity of those bugreports to 'grave', since they cause FTBFS on all the other packages. if those packages used the right library name, they would not be interested by this change. I am wondering why libboost needs to be special-cased to especially supply a '-st' version. Why not force everyone to use '-mt'? special-cased respect who? to not be special case, the single thread would not have the -st and the multi thread would have the -mt. which is the countrary of what you are asking. that is the reason of my change. ciao domenico -[ Domenico Andreoli, aka cavok --[ http://www.dandreoli.com/gpgkey.asc ---[ 3A0F 2F80 F79C 678A 8936 4FEE 0677 9033 A20E BC50 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: [pkg-boost-devel] Bug#429533: dev package changes and its severity
Hi, I've browsed through the bug logs, but this is ugly. yes, it is We're breaking all those APIs, and there is no sane way of finding out what to use. we? who? breaking those APIs? it is only a matter of fixing the already broken library name used to link with programs. i don't see anything broken. So, how is the name of the library name supposed to be 'fixed' ? Also, please keep the severity of those bugreports to 'grave', since they cause FTBFS on all the other packages. if those packages used the right library name, they would not be interested by this change. Please elaborate 'right library name' regards, junichi -- [EMAIL PROTECTED],netfort.gr.jp} Debian Project -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: [pkg-boost-devel] Bug#429533: Bug#429533: dev package changes and its severity
Hi, On Tue, Jun 26, 2007 at 11:25:43PM +0900, Junichi Uekawa wrote: Please elaborate 'right library name' Boost library naming is described in the documentation [0]. Cheers, Domenico [0] http://www.boost.org/more/getting_started/unix-variants.html#library-naming -[ Domenico Andreoli, aka cavok --[ http://www.dandreoli.com/gpgkey.asc ---[ 3A0F 2F80 F79C 678A 8936 4FEE 0677 9033 A20E BC50 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: [pkg-boost-devel] Bug#429533: Bug#429533: dev package changes and its severity
Hi, Please elaborate 'right library name' Boost library naming is described in the documentation [0]. [0] http://www.boost.org/more/getting_started/unix-variants.html#library-naming Which does not mean it is correct for libboost-*-dev to silently change its interface (.so library naming). This description sounds more relevant for cross-platform development and other things, but isn't it possible to provide a saner default for gcc/libc system.s Why not just provide -pthread version (-mt) per default and provide -st version for people who really care? regards, junichi -- [EMAIL PROTECTED],netfort.gr.jp} Debian Project -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: [pkg-boost-devel] Bug#429533: Bug#429533: Bug#429533: dev package changes and its severity
On Wed, Jun 27, 2007 at 05:30:21AM +0900, Junichi Uekawa wrote: Hi, hi, Boost library naming is described in the documentation [0]. [0] http://www.boost.org/more/getting_started/unix-variants.html#library-naming Which does not mean it is correct for libboost-*-dev to silently change its interface (.so library naming). no, silently not. the real bug from my POW is that I did not announced the change. Why not just provide -pthread version (-mt) per default and provide -st version for people who really care? because using -mt versions in place of -st is suboptimal in a way i am not able to evaluate. if building c++ as multi-threaded had a negligible impact, -pthread switch would be the default. moreover careful people is usually the one using multi-threaded stuff, not the opposite. cheers, domenico -[ Domenico Andreoli, aka cavok --[ http://www.dandreoli.com/gpgkey.asc ---[ 3A0F 2F80 F79C 678A 8936 4FEE 0677 9033 A20E BC50 -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#429533: dev package changes and its severity
Hi, I've browsed through the bug logs, but this is ugly. We're breaking all those APIs, and there is no sane way of finding out what to use. Also, please keep the severity of those bugreports to 'grave', since they cause FTBFS on all the other packages. I am wondering why libboost needs to be special-cased to especially supply a '-st' version. Why not force everyone to use '-mt'? regards, junichi -- [EMAIL PROTECTED],debian.org} -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]