On 5/29/07, Andreas Tille <tillea@rki.de> wrote:
On Tue, 29 May 2007, Rudi Cilibrasi, Ph.D. wrote:
> Unfortunately my powerpc mac mini died recently (hard drive crash) and so I have not been able
> to test the powerpc version at all. It may very well work though and should be enabled if so.
It is allways advisable to start with architecture "any" and only replace this
by a list of architectures if serious efforts were done to solve the problem.
I agree. I actually do not know what happened to the powerpc arch. I just mentioned this because
it makes it hard / impossible for me to check and also I did a very large rewrite (everything) between
version 0.9.8 and 0.9.9 so it will be good if somebody tries it (make check). But I wonder whether
the powerpc removal (that I must have done) was accidental or intentional. In the latter case,
it would be for a reason I have now forgotten and we are thus destined to rediscover.
A dying powerpc on your site is no reason at all to remove this architecture
from the list - just let the autobuilders do their work and see what happens.
I am ready to try this on the next rev. Still hoping for any extra info from any interested parties also.
Let me know if I should upload a 0.9.9-2 version to mentors just for the sake of the powerpc addition.
> The only two reasons I can remember that I have removed any architectures are as follows:
> 1) The mips-ish archs have a FTBFS problem related to libpthreads.
This should be sorted out with the porting team or on debian-devel.
I agree and plan to do it as soon as I finish the current task, namely the "big split".
> 2) The 64-bit archs do not support lzmax and ppmdx in particular. I can't remember if powerpc has lzma or ppmd or not... this might
> be the reason or it could simply be an error on my part. If you have such a machine and it seems to work ok, I would like to know.
I would also contact the porting teams. They often have a clue how to work
around these problems and you can make upstream happy with patches that enlarge
the number of architectures their software is working on.
This was my original idea but unfortunately someone filed a bug (#419298) because the lzma and ppmd dependencies can never be satisfied. There is also apparently a good reason that these will never be ported to 64 bit architectures. Basically lzma and ppmd
are no longer maintained and extremely complicated so the chances that anybody will port them is reasonably estimated to be low. So this is the complicated resaon for the complicated Arch line that I would surely like to simplify, correct, and expand as possible as porting allows.
I am glad some people have started using the package and look forward to improving it with your help and feedback. Best regards,
Rudi
--