Remove OSX 10.4 PPC from supported platforms

[Volker Braun]

Two weeks ago I tried to get access to a OSX PPC machine, and we haven't made any progress yet. Thats just not maintainable. I've been playing trac/email tag with Karl-Dieter for two weeks but we still haven't been able to fix the blas linkage. Basically, the OSX PPC binaries that we build are not correctly linked and only work by accident. We can either freeze the current state and never touch the blas again (sucks if you have a modern processor) or we'll have to let go of the ancient stuff. We can still run Sage on OSX PPC as we manage to iron out the kinks, but this shouldn't be a blocker for ATLAS to move forward, say.

[William Stein]

On Monday, September 17, 2012, Volker Braun <vbrau...@gmail.com> wrote:
> Two weeks ago I tried to get access to a OSX PPC machine, and we haven't made any progress yet. Thats just not maintainable. I've been playing trac/email tag with Karl-Dieter for two weeks but we still haven't been able to fix the blas linkage. Basically, the OSX PPC binaries that we build are not correctly linked and only work by accident. We can either freeze the current state and never touch the blas again (sucks if you have a modern processor) or we'll have to let go of the ancient stuff. We can still run Sage on OSX PPC as we manage to iron out the kinks, but this shouldn't be a blocker for ATLAS to move forward, say.
>

+1. We shouldn't hold sage *back* to support 10.4 at this point.

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To post to this group, send email to sage-...@googlegroups.com.
> To unsubscribe from this group, send email to sage-devel+...@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel?hl=en.
>  
>  
>

--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org

[Ivan Andrus]

On Sep 17, 2012, at 4:00 PM, William Stein wrote:

On Monday, September 17, 2012, Volker Braun <vbrau...@gmail.com> wrote:
> Two weeks ago I tried to get access to a OSX PPC machine, and we haven't made any progress yet. Thats just not maintainable. I've been playing trac/email tag with Karl-Dieter for two weeks but we still haven't been able to fix the blas linkage. Basically, the OSX PPC binaries that we build are not correctly linked and only work by accident. We can either freeze the current state and never touch the blas again (sucks if you have a modern processor) or we'll have to let go of the ancient stuff. We can still run Sage on OSX PPC as we manage to iron out the kinks, but this shouldn't be a blocker for ATLAS to move forward, say.
>

+1. We shouldn't hold sage *back* to support 10.4 at this point.

But we should we still try to support 10.4 if it's not too difficult, right?  For example, when building the Mac app, there are certain APIs that appear in 10.5.  I have been able to get by without them, but a few things would be a little easier if I could use them.

-Ivan

[kcrisman]

Since François' idea of a symlink seems to be working, it may be a moot point.  Also, in this particular case the Intel computers seemed to be working fine for some reason.

But the larger point is always a hard one to discuss.  One wants to take advantage of (positive) new things, but in bad cases things can lead to a cycle of planned obsolescence and/or to novelty for its own sake.  I'm sure we've all had a computer sitting around that there was really nothing wrong with, except newer software didn't work on it - but it could still check email and surf the web (maybe even video, with HTML5 now that Flash's reign may be ending).  Or a phone along the same lines!  It would be interesting to quantify how much global demand for certain expensive elements would go down if the product cycle for cell phones was twice as long... not that the Atlas upgrade is causing anything like that!