Bill Hart wrote:
> Ah, we are talking about the same thing I think.
> 
> How silly to pollute the namespace with another macro with the same
> name, especially as MPFR and GMP/MPIR are sister projects. But I am
> guessing this is not an exported symbol but only used internally.
> 
> That means all of my comments regarding the issue with MPN_ZERO are
> irrelevant with respect to David Kirkby's workaround in Sage.
> 
> For various reasons, the correct fix then is precisely what David
> Kirby is doing. That means the sage ticket does need to be reviewed.
> I'll close the MPIR ticket associated with this.
> 
> As for me reviewing the sage ticket, I give up. Too hard for me.
> Someone else can do it please. I don't have time in the foreseeable
> future as it is quite involved. Someone with a working Sage install on
> t2 should do it, as it would be about a day's less work for them than
> it would be for me. Obviously it should be checked that it doesn't
> break the build on some non-T2 machine, e.g. sage.math.
> 
> Bill.

Although Paul's C code is quite simply (only one extra line), the patch 
is more complicated since I wanted to ensure that

1) It is only by default applied on sun4v systems, like 't2', and not on 
any other Solaris version, as it will impact performance.

2) It's possible to override that from an environment variable, so 
binaries could be built on an older (sun4u) system, but allow them to 
run on newer machines like t2, which are sun4v.

To distribute binaries, we should build them on the very first release 
of Solaris 10, otherwise they might fail on systems older than what they 
were built on. To do this we need another machine. I have a relic I can 
set up to do this, but I'd need to force the patch to be applied, just 
in case someone runs it on a system like 't2'


Dave

--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to 
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---

Reply via email to