[crux64] crux64 Digest, Vol 7, Issue 1

Johannes Winkelmann jw at smts.ch
Sun Jun 22 13:55:34 UTC 2008


On Sat, Jun 21, 2008 at 17:55:37 +0200, Bartek Palmowski wrote:
> >
> > Hi
> > > 
> > >     crux64 looks promissing, and I'd me more than happy if I could
> > > help with project. I already have done some initial testing and must
> > > say crux for 64 bit processors is in demand. If I may suggest an idea
> > > to add variable to pkgmk.conf like ie. ARCH_HOST=64, which would make
> > > easier "architecture type" extraction.
> > > 
> >   
> >
> > Why would that be a good idea?
>     Because we could use one ports tree instead of two, which would be 
> easier to maintain.
Having a single tree would mean that if the x86 maintainer would update
a port, it would also be synced to all the other architectures without
any testing.
Also, it would not be possible to have separate versions of the same
port, which can be needed if version X doesn't build/work on some arch.

In addition, some ports are platform specific (e.g. lilo (x86), silo
(sparc) etc.), and it wouldn't make sense to sync them to the user.

Another big drawback of any "single tree" approach is that you need
commit access to that tree. A separated approach means that anyone can
start with a new architecture, without being dependent on others.

Finally, I don't really buy the "easier to maintain" argument. The hard
part is to get a port working. That's an initial effort. When you then
do an update of that port, most time is spent on building and testing
it, not bumping the version in Pkgfile. That's independent from whether
there's a single tree, or separate ones.

Hope this explains it,
Regards Johannes
Johannes Winkelmann              mailto:jw at smts.ch
Zurich, Switzerland              http://jw.smts.ch

More information about the crux64 mailing list