Fwd: Sage 5.2 fails to build (mpir-2.4.0.p6) on Fedora 16 32-bit

[Rick Graham]

Hello,

After failing an upgrade from Sage v. 4.8, the Sage v. 5.2 source was
downloaded (md5sum checked fine) and the build started. The build
failed with mpir-2.4.0.p6 complaining that "configure: error: ABI=32
is not among the following valid choices: standard".

The config.log and mpir-2.4.0.p6.log are attached.

Thanks for your help.
Rick

[rickhg12hs]

On Monday, September 10, 2012 2:16:02 PM UTC-4, rickhg12hs wrote:

After failing an upgrade from Sage v. 4.8, the Sage v. 5.2 source was
downloaded (md5sum checked fine) and the build started.  The build
failed with mpir-2.4.0.p6 complaining that "configure: error: ABI=32
is not among the following valid choices: standard".

I ran "make -k" just so the build would continue after the mpir failure and another problem was encountered.  Something needing Fedora 16's openssl-devel failed and after installing it from the normal Fedora repo the only build failure now seems to be mpir.  Fedora 16's mpir is several versions older than Sage's 5.2 so I'm wondering if there is just another dependency I need to successfully build Sage 5.2.

Any thoughts?

[rickhg12hs]

On Monday, September 10, 2012 2:16:02 PM UTC-4, rickhg12hs wrote:

After failing an upgrade from Sage v. 4.8, the Sage v. 5.2 source was
downloaded (md5sum checked fine) and the build started.  The build
failed with mpir-2.4.0.p6 complaining that "configure: error: ABI=32
is not among the following valid choices: standard".

Not done testing yet, and it's not ideal, but this workaround built completely without error.

$ MPIR_CONFIGURE='ABI=standard' make

[rickhg12hs]

Assigning  MPIR_CONFIGURE as above before executing make does seem to be an effective workaround.  Everything was built and all the tests (./sage -testall) passed.

Perhaps there's a configuration detection problem that requires this workaround?