I placed my test boost-1.33 srpm here: http://nbecker.dyndns.org:8080/boost-1.33.0-6.src.rpm Basically, I modified the redhat 1.32 spec file: 1) Don't mess with soname 2) Autoconfigure icu support 3) Build all variants 4) Add my (little) patch for serializing mersenne_twister. Yes, I know it's not optimal, but it's simple and good enough for my purposes. Please comment.
Neal Becker wrote:
I placed my test boost-1.33 srpm here:
That's great, thanks a lot !
Basically, I modified the redhat 1.32 spec file:
1) Don't mess with soname 2) Autoconfigure icu support
Will there be a central authority that distributes 'official' rpms ? Will it depend (and use) icu or not ? I'm just afraid that at some point multiple conflicting rpm versions will be available and confuse potential users.
3) Build all variants
Given that boost libs are quite big (in particular the ones with debug info), I'd suggest to provide separate packages containing individual variants, so users can install only the one they really want. Thanks for your efford ! Stefan
Stefan Seefeld wrote:
Neal Becker wrote:
I placed my test boost-1.33 srpm here:
That's great, thanks a lot !
Basically, I modified the redhat 1.32 spec file:
1) Don't mess with soname 2) Autoconfigure icu support
Will there be a central authority that distributes 'official' rpms ? Will it depend (and use) icu or not ? I'm just afraid that at some point multiple conflicting rpm versions will be available and confuse potential users.
I don't know. If I can get something together that boost.devel readers generally approve, I will try to push it to Redhat. In the past, I have not had good cooperation, but we'll see.
3) Build all variants
Given that boost libs are quite big (in particular the ones with debug info), I'd suggest to provide separate packages containing individual variants, so users can install only the one they really want.
Good idea. I think probably make debug version separate. I don't know how to do that, though. rpmbuild automatically creates -debug versions, so I'll have to learn how I can control that process. I don't think separate -mt version is needed - that would be inconsistent with other common libs.
On Aug 12, 2005, at 10:38 AM, Neal Becker wrote:
Stefan Seefeld wrote:
Will there be a central authority that distributes 'official' rpms ? Will it depend (and use) icu or not ? I'm just afraid that at some point multiple conflicting rpm versions will be available and confuse potential users.
I don't know. If I can get something together that boost.devel readers generally approve, I will try to push it to Redhat. In the past, I have not had good cooperation, but we'll see.
We can also make it available through Sourceforge, along with the other archives of 1.33.0. Doug
participants (3)
-
Douglas Gregor -
Neal Becker -
Stefan Seefeld