Re: Fortran in the base
Hasso Tepper <[EMAIL PROTECTED]> wrote: > Simon 'corecode' Schubert wrote: >> we don't have objc support in gcc41, but nobody complained since i >> imported/switched the default version, so I figured nobody uses it. If >> you do, please speak up! (And if possible add the support) > > The user who mailed to me privately about objc is using gcc34 by default. > I know several more users (wasn't Rumko one of them?) who do that because > of the issues they have with various software pieces - qemu and festival > are two pieces I have seen compile/work just fine with gcc-3.x, but blow > up with gcc-4.x. I still use gcc34 for buildworld, as the va_copy() problem http://leaf.dragonflybsd.org/mailarchive/users/2007-11/msg00032.html still exists with gcc41 built libc. Johannes
Re: Fortran in the base
Hasso Tepper wrote: > Simon 'corecode' Schubert wrote: >> we don't have objc support in gcc41, but nobody complained since i >> imported/switched the default version, so I figured nobody uses it. If >> you do, please speak up! (And if possible add the support) > > The user who mailed to me privately about objc is using gcc34 by default. > I know several more users (wasn't Rumko one of them?) who do that because > of the issues they have with various software pieces - qemu and festival > are two pieces I have seen compile/work just fine with gcc-3.x, but blow > up with gcc-4.x. > > I will look at bringing objc support into gcc41 later. > > I don't use gcc34 by default, I only use it specifically for qemu (I manually set CCVER when I'm off to compile qemu). -- Regards, Rumko
Re: Fortran in the base
Simon 'corecode' Schubert wrote: > we don't have objc support in gcc41, but nobody complained since i > imported/switched the default version, so I figured nobody uses it. If > you do, please speak up! (And if possible add the support) The user who mailed to me privately about objc is using gcc34 by default. I know several more users (wasn't Rumko one of them?) who do that because of the issues they have with various software pieces - qemu and festival are two pieces I have seen compile/work just fine with gcc-3.x, but blow up with gcc-4.x. I will look at bringing objc support into gcc41 later. -- Hasso Tepper
Re: Fortran in the base
Hasso Tepper wrote: > Hasso Tepper wrote: >> And while at looking what needs to removed I discovered that we have >> Objc support as well in gcc34. It's compiled by default, but we don't >> have even /usr/bin/cc1objc. Any objections to remove it as well? > > Nevermind. I received already one mail from user that he actually uses > Objective C. And as we don't have alternative to that from pkgsrc (unless > someone will make gcc version(s) in pkgsrc work in DragonFly), we should > keep it. It's not that much of code. we don't have objc support in gcc41, but nobody complained since i imported/switched the default version, so I figured nobody uses it. If you do, please speak up! (And if possible add the support) cheers simon signature.asc Description: OpenPGP digital signature
Re: Fortran in the base
Hasso Tepper wrote: > And while at looking what needs to removed I discovered that we have > Objc support as well in gcc34. It's compiled by default, but we don't > have even /usr/bin/cc1objc. Any objections to remove it as well? Nevermind. I received already one mail from user that he actually uses Objective C. And as we don't have alternative to that from pkgsrc (unless someone will make gcc version(s) in pkgsrc work in DragonFly), we should keep it. It's not that much of code. -- Hasso Tepper
Re: Fortran in the base
:Simon 'corecode' Schubert wrote: :> Hasso Tepper wrote: :> > OK, much clearer now. So, now I think that best approach is :> > completely opposite. Basically, there are two options: :> > :> > a) Remove fortran from the base. :> :> +1 if f2c works well. : :My brief experience shows that it does. : :And while at looking what needs to removed I discovered that we have Objc :support as well in gcc34. It's compiled by default, but we don't have :even /usr/bin/cc1objc. Any objections to remove it as well? : :-- :Hasso Tepper It sounds good to me. My perspective on pkgsrc is that if we can do anything to make pkgsrc operate better, we should. -Matt Matthew Dillon <[EMAIL PROTECTED]>
Re: Fortran in the base
Simon 'corecode' Schubert wrote: > Hasso Tepper wrote: > > OK, much clearer now. So, now I think that best approach is > > completely opposite. Basically, there are two options: > > > > a) Remove fortran from the base. > > +1 if f2c works well. My brief experience shows that it does. And while at looking what needs to removed I discovered that we have Objc support as well in gcc34. It's compiled by default, but we don't have even /usr/bin/cc1objc. Any objections to remove it as well? -- Hasso Tepper
Re: Fortran in the base
Hasso Tepper wrote: > OK, much clearer now. So, now I think that best approach is completely > opposite. Basically, there are two options: > > a) Remove fortran from the base. +1 if f2c works well. cheers simon signature.asc Description: OpenPGP digital signature
Re: Fortran in the base
Rumko wrote: > Well ... the biggest problem isn't that we don't have fortran in our > gcc-4.1 but that pkgsrc checks for fortran only by checking if the > files exist not by actually running/testing if it actually works (try > running g77 -v or f77 -v without setting CCVER to gcc34) afaik (if i > remember correctly, deleting f77&co. makes pkgsrc automatically pull in > f2c so that everything after that works, but it's been quite some time > since i last tried that, so i could be remembering it incorrectly). > You can override the problems with fortran by adding "PKGSRC_COMPILER= > f2c gcc" in your mk.conf (or was it PKGSRC_COMPILERS? would have to > check). But since it seems that pkgsrc has broken checks regarding > ONLY_FOR_COMPILER you won't be able to compile packages such as mplayer > after adding that line in your mk.conf (I usually compile mplayer > manually by commenting that line in mk.conf after everything else is > done). OK, much clearer now. So, now I think that best approach is completely opposite. Basically, there are two options: a) Remove fortran from the base. b) Make NO_FORTRAN default for now. And fortran will be removed from the base if we'll remove gcc34 somewhere in future. AFAICS NetBSD (at least recent versions) don't have fortran in the base either and rely on lang/f2c. Opinions? -- Hasso Tepper
Re: Fortran in the base
Hasso Tepper wrote: The question is what we should do? Due to fact that it's quite widely used, we should do something, but what? I'd prefer to add fortran into gcc-4.1 as well. Are there objections? If you want to add it, I wouldn't object. Sascha -- http://yoyodyne.ath.cx
Re: Fortran in the base
Hasso Tepper wrote: > Quite a big number of pkgsrc build failures come from fact that we don't > have Fortran compiler in our gcc-4.1. We have it in gcc-3.4 though, but > for some reason not in gcc-4.1. > > The question is what we should do? Due to fact that it's quite widely > used, we should do something, but what? I'd prefer to add fortran into > gcc-4.1 as well. Are there objections? > > Well ... the biggest problem isn't that we don't have fortran in our gcc-4.1 but that pkgsrc checks for fortran only by checking if the files exist not by actually running/testing if it actually works (try running g77 -v or f77 -v without setting CCVER to gcc34) afaik (if i remember correctly, deleting f77&co. makes pkgsrc automatically pull in f2c so that everything after that works, but it's been quite some time since i last tried that, so i could be remembering it incorrectly). You can override the problems with fortran by adding "PKGSRC_COMPILER= f2c gcc" in your mk.conf (or was it PKGSRC_COMPILERS? would have to check). But since it seems that pkgsrc has broken checks regarding ONLY_FOR_COMPILER you won't be able to compile packages such as mplayer after adding that line in your mk.conf (I usually compile mplayer manually by commenting that line in mk.conf after everything else is done). my 2 cents -- Regards, Rumko
Fortran in the base
Quite a big number of pkgsrc build failures come from fact that we don't have Fortran compiler in our gcc-4.1. We have it in gcc-3.4 though, but for some reason not in gcc-4.1. The question is what we should do? Due to fact that it's quite widely used, we should do something, but what? I'd prefer to add fortran into gcc-4.1 as well. Are there objections? -- Hasso Tepper