Rob, I've committed a fix for this and it now seems to work on Mac OS X. It turned out the problem was that the code of the module was being loaded into the normal heap which no longer has execute permission rather than the code area. It looks as though Linux, unlike Mac OS X and Windows, ignores the lack of execute permission.
The module name has also changed to IntInfAsInt.
Regards, David
On 05/03/2017 20:46, Rob Arthan wrote:
David,
On 5 Mar 2017, at 19:23, David Matthews <David.Matthews at prolingua.co.uk> wrote:
On 04/03/2017 16:42, Rob Arthan wrote:
I?ve just pulled the latest source. I am experimenting with how to build ProofPower if the installed Poly/ML has fixed magnitude integers. I get a bus error when I do:
PolyML.loadModule "modules/IntAsIntInf/IntAsIntInf?;
I've just tried this. I was getting an infinite loop rather than a bus error and only on 64-bits. I've pushed a fix for that. Has it sorted out the bus error?
I should have said that I was doing my experiments on Mac OS (Sierra 10.12.3). I?ve just tried it on a Ubuntu 16.04 VirtualBox VM and there I see the same behaviour as you: infinite loop before your fix and OK after. On Mac OS I am still getting the bus error:
rda]- ./poly Poly/ML 5.6.1 Testing (Git version v5.6-835-geec20d3)
PolyML.loadModule "modules/IntAsIntInf/IntAsIntInf" ;
Bus error: 10
I should say that this is not a stopper for me as I am currently planning to accommodate the choice that was made when Poly/ML was built rather than try to override it with the module.
(By the way, why is the configure option called ?enable-intinf-as-int but it?s the other way round in the module name?)
That is inconsistent. I'm not sure which is better.
Nor me! If you want to make it consistent, I suggest you change whichever is least work.
Regards,
Rob.