I don't see that myself but it's certainly possible. A delay followed by Error 1 suggests that it is relying on the crow-bar thread to stop. The fact that this only happens on some platforms suggests a race condition.
In view of this I'm inclined to release the version without this change (44b7b88) as 5.7.1 and investigate the problem later. Are you certain the problem you've identified doesn't happen in that version?
David
On 25/11/2017 21:45, Makarius wrote:
On 25/11/17 18:44, David Matthews wrote:
I've had a better look and I found that I was seeing this as well.? I've pushed a fix and it no longer seems to be doing it.? It's a very small change so I would be very surprised if it has broken anything but I'll give it a couple of days and then release 5.7.1.
This change a24f39a95ca9 causes a new problem when building modules/IntInfAsInt:
... echo "use "./ROOT.sml";" | ../../poly -q -error-exit Error 1
Before emitting that error, it appears to "hang" for several seconds.
The build procedure is that of Isabelle, e.g.:
$ isabelle build_polyml -m32 -s sha1 polyml-git --with-gmp $ isabelle build_polyml -m64 -s sha1 polyml-git --with-gmp
I've also seen the error with manual invocation of:
./configure --enable-intinf-as-int && make clean && make compiler
This effect is somewhat non-deterministic, but it appears quite reliably on my reference machines for x86-linux and x86_64-linux (Ubuntu 12.04 LTS).
In most other system configurations it appears to work.
Makarius _______________________________________________ polyml mailing list polyml at inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/polyml