8.2.rc2 on OSX: compile gcc 7.2 after git pull

[David....@inria.fr]

I just made a "git pull" to go from 8.2.rc1 to 8.2.rc2 on my OSX laptop, and "make" launches the compilation of gcc 7.2 ?!?

I had to do a "make dist-clean" for 8.2.beta7, since my install missed the switch from gcc to clang on osx (https://groups.google.com/forum/#!topic/sage-devel/xP7F5T350IQ).
So I assume that my installation is now using clang.
I don't understand why the last git pull forces to switch back to gcc.

Should I do a dist-clean again ? This is boring since I will then have to re-install multiple optional packages + Cplex + Gurobi + etc. but apparently I will have no other choice.

Any intel is more than welcome.
David.

[François Bissey]

There was trouble with the latest Xcode release from Apple. It broke gfan in subtle ways
which translated in a number of doctests failures. So we decided to temporarily switch back
to gcc for the imminent release. We should have communicated things better on sage-release.
In the last 24hours Dima found the problem and we have a fix so there may still be a chance
of 8.2 using clang out of the box.

François

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

[Dima Pasechnik]

I suggest you ignore rc2 for the time being. The (too quick, IMHO - I had trouble reproducing this on a remotely 

accessible OSX box, and had to get back home to an old Apple laptop) change back to gcc was due to #25113, for which

we now have a workaround on https://trac.sagemath.org/ticket/25118

The latter will be turned into a proper branch later today.

Cheers,

Dima

David.

[Dima Pasechnik]

Perhaps we need a howto page on wiki on OSX, and more generally, clang, debugging - how to turn core dumping on, how to run LLVM debugger lldb...

And most probably an analogue of gdb package for clang, too, is needed. 

[Dima Pasechnik]

https://trac.sagemath.org/ticket/25118 is ready for review now; in particular it reverts the change back to gcc, so staying with clang,

and fixes the crashes on OSX from #25113.

On Thursday, April 12, 2018 at 8:46:35 AM UTC+1, David....@inria.fr wrote:

[Volker Braun]

On Thursday, April 12, 2018 at 10:13:50 AM UTC+2, François Bissey wrote:

In the last 24hours Dima found the problem and we have a fix so there may still be a chance
of 8.2 using clang out of the box. 

IMHO the safe bet is to use gcc in 8.2 (as it was in 8.1) instead of last-minute experiments (how many different macs did you test this on?)

[François Bissey]

Between Dima and me: 2

[Dima Pasechnik]

3 - also a Xeon at LRI (which is unaffected by this, but still...)

 

[Dima Pasechnik]

The fact that gcc is not miscompiled by the new Xcode is also sort of last-minute, isn't it?

And just for the time for 8.3 we will get Xcode 6.4 (it already exists in beta), and so the work we did in the past 1.5 years porting Sage to native Xcode will never see the light of the day, right?

IMHO it's OK if the release takes a day or two more, really...

 

[David Roe]

What kinds of Macs would be useful to test it on?

David

 

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.

To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscribe@googlegroups.com.

[Dima Pasechnik]

ones that have xcode 6.3.

it appears that the bug (hopefully) fixed by #25118 shows on all CPUs, except Xeons (the latter not used in laptops afaik, only in desktops etc).

[David....@inria.fr]

I will then try to get back to rc1.
Thank you all.
David.