Robert Bradshaw wrote:
On Feb 21, 2010, at 12:53 PM, Dr. David Kirkby wrote:
The addition of
http://trac.sagemath.org/sage_trac/ticket/6583 (implement 2-isogeny
descent over QQ natively in Sage using ratpoints)
sage-4.3.1 broke the build of Sage on Solaris. I suspect this was the
ticket, but Minh had proved this some time back, after starting with
Sage 4.3, he added all the patches until the Sage build broke. It was
#6583 which broke the build.
(Since then, someone updated R without checking that on Solaris, but
that issue has been resolved - or at least I believe it has been
resolved. I noticed Minh attempted a build of sage-4.3.3.alpha1 on t2,
using my updated iconv and R packages. Both iconv and R built OK, but
the Sage library is failing.
I've created a trac ticket for this bug
http://trac.sagemath.org/sage_trac/ticket/7867
Robert Miller shows on this ticket what is causing the problem.
/*
* It is invalid to compile an XPG3, XPG4, XPG4v2, or XPG5 application
* using c99. The same is true for POSIX.1-1990, POSIX.2-1992, POSIX.1b,
* and POSIX.1c applications. Likewise, it is invalid to compile an XPG6
* or a POSIX.1-2001 application with anything other than a c99 or later
* compiler. Therefore, we force an error in both cases.
*/
and suspected it might be related to the fact the ratpoints code is old.
Assuming the R (#8285) and iconv (#8191) tickets get positive review
(and I note Minh is going to look at them), it only leaves this issue
that is stopping Sage building on Solaris 10 (SPARC).
Does anyone have any ideas how to get around this?
I don't know how important #6583 it so Sage, or how important William
sees a Solaris port as important, but at the minute, these two appear
to be mutually exclusive.
I don't have any instant fixes, but can confirm that #6583 is very
important to some of us.
- Robert
I appreciate that, but would also appreciate if William would comment where he
stands on this.
He sent me an email only a couple of weeks ago.
===============================================================
David,
(1) I couldn't get anywhere building Sage on x86 Solaris on skynet (fulvia).
Can you? This was pretty annoying to the people that bought us fulvia.
(2) Sun wants to know if we have a Sage available yet for t2. See below.
I really need to shift into the mode of actually providing something that
*works* on Solaris, despite hickups, rather than just polishing foundations...
(I've removed the rest, as it is potentially confidential)
=================================================================
So unless a fix is found, a decision needs to be made whether we have a Solaris
port of Sage whether we have patch #6583 in Sage.
Could those that need #6583 not install that manually, or include it as an
optional package?
William has taken free hardware from Sun (t2) and others bought OpenSolaris
hardware for Sage development. Micheal was paid full-time to work on a Solaris
port. This is all going to be wasted if the port is let to stagnate, which is
what will happen very soon I feel.
I know we have a working 4.3.0.1, but the fact it is old means new patches can't
be tested against it.
Whilst the SPARC port is tied up by this patch, I personally find it hard to get
any motivation to continue the OpenSolaris port, knowing there are problems in
the library like this.
Dave
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
