Re: [gentoo-dev] Re: [RFC] New policy: LDFLAGS should be respected
On L, 2008-07-26 at 03:39 +0300, Nikos Chantziaras wrote: > Fortunately, the majority of ebuilds/packages are honoring LDFLAGS. Of > course it's kinda difficult to always check if a package honors it or > not. But it's a good idea to file a bug for every package that does not > honor it (without a reason). I guess as many are using it to pass --hash-style=gnu in addition to other things[1], an easy way to find out which don't honor it out of your installed packages is to scan for ELF files that contain the .hash ELF section in addition to .gnu.hash ELF section. Something through scanelf -q -k .hash outputting anything or not could be used to determine if the section exists or not. I'm sure there are better ways. This doesn't work for packages that don't ship ELF files, but ld works on ELF files, so those that don't use ELF files shouldn't really care if LDFLAGS is honored or not... Maybe this gives some ideas to someone to write a proper QA script, or point us all to an already existing almighty script or tool that does just that. -- Mart Raudsepp Gentoo Developer Mail: [EMAIL PROTECTED] Weblog: http://planet.gentoo.org/developers/leio signature.asc Description: This is a digitally signed message part
Re: [gentoo-dev] Re: [RFC] New policy: LDFLAGS should be respected
On Sat, 26 Jul 2008 00:15:03 + (UTC) Duncan <[EMAIL PROTECTED]> wrote: > In particular, --as-needed makes a HUGE very practical difference. > It may or may not be the wrong answer to the problem in theory, but > lacking anything even close to as workable right now, that alone is > IMO reason enough to work to get LDFLAGS honored. I appreciate the > difference it made here every time I run revdep-rebuild! Ignoring CFLAGS on some archs results in code that is either an order of magnitude slower or just plain won't run. Ignoring LDFLAGS means on those rare occasions when libraries aren't slotted properly you have to rebuild a few more things. Rather a large difference in impact there... -- Ciaran McCreesh signature.asc Description: PGP signature
[gentoo-dev] Re: [RFC] New policy: LDFLAGS should be respected
Duncan wrote: Nikos Chantziaras <[EMAIL PROTECTED]> posted [EMAIL PROTECTED], excerpted below, on Sat, 26 Jul 2008 02:54:07 +0300: Gilles Dartiguelongue wrote: Le jeudi 24 juillet 2008 à 18:36 +0200, Arfrever Frehtes Taifersar Arahesis a écrit : I would like to suggest new policy stating that packages should respect LDFLAGS. This policy is required to allow QA team to fix packages which ignore LDFLAGS. this question might sound stupid, but what are you actually trying to fix ? What do these packages do or do not do by ignoring your ldflags that is so crucial to you ? "-Wl,-O1,--hash-style=gnu,--sort-common,--as-needed" In particular, --as-needed makes a HUGE very practical difference. It may or may not be the wrong answer to the problem in theory, but lacking anything even close to as workable right now, that alone is IMO reason enough to work to get LDFLAGS honored. I appreciate the difference it made here every time I run revdep-rebuild! That's what makes observation of LDFLAGS very practically critical to me. Fortunately, the majority of ebuilds/packages are honoring LDFLAGS. Of course it's kinda difficult to always check if a package honors it or not. But it's a good idea to file a bug for every package that does not honor it (without a reason).
[gentoo-dev] Re: [RFC] New policy: LDFLAGS should be respected
Nikos Chantziaras <[EMAIL PROTECTED]> posted [EMAIL PROTECTED], excerpted below, on Sat, 26 Jul 2008 02:54:07 +0300: > Gilles Dartiguelongue wrote: >> Le jeudi 24 juillet 2008 à 18:36 +0200, Arfrever Frehtes Taifersar >> Arahesis a écrit : >>> I would like to suggest new policy stating that packages should >>> respect LDFLAGS. >>> This policy is required to allow QA team to fix packages which ignore >>> LDFLAGS. >>> >> this question might sound stupid, but what are you actually trying to >> fix ? What do these packages do or do not do by ignoring your ldflags >> that is so crucial to you ? > > "-Wl,-O1,--hash-style=gnu,--sort-common,--as-needed" In particular, --as-needed makes a HUGE very practical difference. It may or may not be the wrong answer to the problem in theory, but lacking anything even close to as workable right now, that alone is IMO reason enough to work to get LDFLAGS honored. I appreciate the difference it made here every time I run revdep-rebuild! That's what makes observation of LDFLAGS very practically critical to me. -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman
[gentoo-dev] Re: [RFC] New policy: LDFLAGS should be respected
Gilles Dartiguelongue wrote: Le jeudi 24 juillet 2008 à 18:36 +0200, Arfrever Frehtes Taifersar Arahesis a écrit : I would like to suggest new policy stating that packages should respect LDFLAGS. Small amount of packages which ignore LDFLAGS should be patched to respect them. Such patches are usually small and easy to write. This policy is required to allow QA team to fix packages which ignore LDFLAGS. this question might sound stupid, but what are you actually trying to fix ? What do these packages do or do not do by ignoring your ldflags that is so crucial to you ? "-Wl,-O1,--hash-style=gnu,--sort-common,--as-needed"
Re: [gentoo-dev] [RFC] New policy: LDFLAGS should be respected
Le jeudi 24 juillet 2008 à 18:36 +0200, Arfrever Frehtes Taifersar Arahesis a écrit : > I would like to suggest new policy stating that packages should respect > LDFLAGS. > Small amount of packages which ignore LDFLAGS should be patched to respect > them. > Such patches are usually small and easy to write. > > This policy is required to allow QA team to fix packages which ignore LDFLAGS. > this question might sound stupid, but what are you actually trying to fix ? What do these packages do or do not do by ignoring your ldflags that is so crucial to you ? -- Gilles Dartiguelongue <[EMAIL PROTECTED]> Gentoo signature.asc Description: Ceci est une partie de message numériquement signée
Re: [gentoo-dev] SDLMame maintainer with Gnome setup wanted
Christian Birchinger wrote: Hello Anyone interested in maintaining further SDLMame updates? Beginning with 0.126 it requires GConf to get a font setting for it's now mandatory debugger. I use a plain XFCE setup and don't really want to install stuff like Orbit and GConf etc. My patch to have GConf optional was rejected by upstream: https://bugs.gentoo.org/show_bug.cgi?id=232233 http://www.bannister.org/forums/ubbthreads.php?ubb=showflat&Number=42698&page=3 seemed that it would be acceptable from what I read, _maybe_ I'd check if would be possible use fontconfig directly. lu -- Luca Barbato Gentoo Council Member Gentoo/linux Gentoo/PPC http://dev.gentoo.org/~lu_zero
Re: [gentoo-dev] SDLMame maintainer with Gnome setup wanted
Rémi Cardona wrote: Christian Birchinger a écrit : I use a plain XFCE setup and don't really want to install stuff like Orbit and GConf etc. FWIW, upcoming versions of GConf will use dbus instead of orbit for IPC. That should reduce all the trouble we've all experienced with corba/orbit. gconf is still problematic nonetheless. -- Luca Barbato Gentoo Council Member Gentoo/linux Gentoo/PPC http://dev.gentoo.org/~lu_zero
Re: [gentoo-dev] RFC: Should 'imlib' USE flag be on by default in x86 and amd64 desktop profiles?
On Fri, 2008-07-25 at 15:00 -0400, Jim Ramsay wrote: > I've run into it a few times now that fluxbox users running Gentoo > wonder why they can't get icons to work in the fluxbox menus. The > short answer is that 'imlib' is off by default in many profiles, > including default-linux/amd64/2007.0/desktop and > default-linux/x86/2007.0/desktop > > I know that I could turn it on by default for fluxbox only using EAPI-1, > but since it's a global USE flag, the profiles may be a better place. > > I think imlib is something most desktop users would want, since it lets > them see all those pretty graphics. Comments? Concerns? imlib is unmaintained/deprecated. So I don't think it makes sense to push it by default to users using GNOME/KDE at least (because they have better image loading libraries). -- Olivier Crête [EMAIL PROTECTED] Gentoo Developer
Re: [gentoo-dev] RFC: Should 'imlib' USE flag be on by default in x86 and amd64 desktop profiles?
Jim Ramsay kirjoitti: I've run into it a few times now that fluxbox users running Gentoo wonder why they can't get icons to work in the fluxbox menus. The short answer is that 'imlib' is off by default in many profiles, including default-linux/amd64/2007.0/desktop and default-linux/x86/2007.0/desktop I know that I could turn it on by default for fluxbox only using EAPI-1, but since it's a global USE flag, the profiles may be a better place. I think imlib is something most desktop users would want, since it lets them see all those pretty graphics. Comments? Concerns? I run qgrep -He "IUSE.*imlib[^2]" to see where it's used. I think it doesn't make sense to enable imlib by default for at least ffmpeg. But then the question remains whether it should be on by default and then turned of where it doesn't make sense. IMHO you should put in EAPI 1 defaults already and then remove them if it gets enabled globally. Regards, Petteri signature.asc Description: OpenPGP digital signature
[gentoo-dev] RFC: Should 'imlib' USE flag be on by default in x86 and amd64 desktop profiles?
I've run into it a few times now that fluxbox users running Gentoo wonder why they can't get icons to work in the fluxbox menus. The short answer is that 'imlib' is off by default in many profiles, including default-linux/amd64/2007.0/desktop and default-linux/x86/2007.0/desktop I know that I could turn it on by default for fluxbox only using EAPI-1, but since it's a global USE flag, the profiles may be a better place. I think imlib is something most desktop users would want, since it lets them see all those pretty graphics. Comments? Concerns? -- Jim Ramsay Gentoo/Linux Developer (rox/gkrellm/fluxbox) signature.asc Description: PGP signature
Re: [gentoo-dev] SDLMame maintainer with Gnome setup wanted
Christian Birchinger a écrit : I use a plain XFCE setup and don't really want to install stuff like Orbit and GConf etc. FWIW, upcoming versions of GConf will use dbus instead of orbit for IPC. That should reduce all the trouble we've all experienced with corba/orbit. Cheers, Rémi