Since I haven't had any recent bug reports I've uploaded the current SVN snap-shot to SourceForge as version 5.5.2.
Most of the changes are bug fixes. One major fix was to the intermediate code optimiser reported back in September. Other changes are the addition of PrettyStringWithWidth and PrettyLineBreak constructors to the PolyML.pretty datatype, Thread.Thread.numPhysicalProcessors to return the number of physical processors and printing ?.t in circumstances where t is bound to a different type. The debugger has been extended to include structures and the arguments to functors.
As always, let me know if you find bugs or problems.
David
I just tried building from 5.5.2-fixes svn and it seems like no libpolymain.a was generated - has that been removed, or did I do something wrong? I did configure with --enable-shared.
On Fri, May 9, 2014 at 4:00 PM, David Matthews < David.Matthews at prolingua.co.uk> wrote:
Since I haven't had any recent bug reports I've uploaded the current SVN snap-shot to SourceForge as version 5.5.2.
Most of the changes are bug fixes. One major fix was to the intermediate code optimiser reported back in September. Other changes are the addition of PrettyStringWithWidth and PrettyLineBreak constructors to the PolyML.pretty datatype, Thread.Thread.numPhysicalProcessors to return the number of physical processors and printing ?.t in circumstances where t is bound to a different type. The debugger has been extended to include structures and the arguments to functors.
As always, let me know if you find bugs or problems.
David _______________________________________________ polyml mailing list polyml at inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/polyml
Sorry for the noise - I found the problem which was related to Arch Linux's build system stripping away the .a files by default. I've set the switch for it not to do that now. If anyone wants to use Poly/ML on Arch Linux or a derivative, the package is available here: https://aur.archlinux.org/packages/polyml-svn/
On Fri, May 9, 2014 at 5:22 PM, Ramana Kumar <Ramana.Kumar at cl.cam.ac.uk>wrote:
I just tried building from 5.5.2-fixes svn and it seems like no libpolymain.a was generated - has that been removed, or did I do something wrong? I did configure with --enable-shared.
On Fri, May 9, 2014 at 4:00 PM, David Matthews < David.Matthews at prolingua.co.uk> wrote:
Since I haven't had any recent bug reports I've uploaded the current SVN snap-shot to SourceForge as version 5.5.2.
Most of the changes are bug fixes. One major fix was to the intermediate code optimiser reported back in September. Other changes are the addition of PrettyStringWithWidth and PrettyLineBreak constructors to the PolyML.pretty datatype, Thread.Thread.numPhysicalProcessors to return the number of physical processors and printing ?.t in circumstances where t is bound to a different type. The debugger has been extended to include structures and the arguments to functors.
As always, let me know if you find bugs or problems.
David _______________________________________________ polyml mailing list polyml at inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/polyml
On 9 May 2014, at 16:00, David Matthews <David.Matthews at prolingua.co.uk> wrote:
Since I haven't had any recent bug reports I've uploaded the current SVN snap-shot to SourceForge as version 5.5.2.
Most of the changes are bug fixes. One major fix was to the intermediate code optimiser reported back in September. Other changes are the addition of PrettyStringWithWidth and PrettyLineBreak constructors to the PolyML.pretty datatype, Thread.Thread.numPhysicalProcessors to return the number of physical processors and printing ?.t in circumstances where t is bound to a different type. The debugger has been extended to include structures and the arguments to functors.
As always, let me know if you find bugs or problems.
The line that sets EXTRALDFLAGS for Mac OS X in polyc.in is missing a double quote character and this causes polyc to raise a syntax error.
When I fixed polyc, it gave the warning that was discussed in an earlier thread about version 5.5.1 (http://lists.inf.ed.ac.uk/pipermail/polyml/2013-September/001353.html):
ld: warning: could not create compact unwind for _ffi_call_unix64: does not use RBP or RSP based frame
This doesn?t seem to have any adverse effects.
Regards,
Rob.
On 10/05/2014 15:07, Rob Arthan wrote:
The line that sets EXTRALDFLAGS for Mac OS X in polyc.in is missing a double quote character and this causes polyc to raise a syntax error.
I've fixed that in SVN trunk and the fixes-5.2.2 branch. I wonder whether it is worth exporting that and updating the tar.gz file on SourceForge.
When I fixed polyc, it gave the warning that was discussed in an earlier thread about version 5.5.1 (http://lists.inf.ed.ac.uk/pipermail/polyml/2013-September/001353.html):
ld: warning: could not create compact unwind for _ffi_call_unix64: does not use RBP or RSP based frame
This doesn?t seem to have any adverse effects.
I did look into this and committed a "fix": adding -no-compact-unwind. Unfortunately this broke C++ exception handling in Mac OS X Mavericks so I reverted it. Since it seems to be just a warning that seems preferable to having poly crash every time the RTS raised an exception.
David
David,
On 11 May 2014, at 12:23, David Matthews <David.Matthews at prolingua.co.uk> wrote:
On 10/05/2014 15:07, Rob Arthan wrote:
The line that sets EXTRALDFLAGS for Mac OS X in polyc.in is missing a double quote character and this causes polyc to raise a syntax error.
I've fixed that in SVN trunk and the fixes-5.2.2 branch. I wonder whether it is worth exporting that and updating the tar.gz file on SourceForge.
I think it might be worth fixing the 5.5.2 release tar.gz file as you have been encouraging people to use polyc and without the fix it just doesn?t work at all on Mac OS X.
Regards,
Rob.
On 11/05/2014 17:21, Rob Arthan wrote:
David,
On 11 May 2014, at 12:23, David Matthews <David.Matthews at prolingua.co.uk> wrote:
On 10/05/2014 15:07, Rob Arthan wrote:
The line that sets EXTRALDFLAGS for Mac OS X in polyc.in is missing a double quote character and this causes polyc to raise a syntax error.
I've fixed that in SVN trunk and the fixes-5.2.2 branch. I wonder whether it is worth exporting that and updating the tar.gz file on SourceForge.
I think it might be worth fixing the 5.5.2 release tar.gz file as you have been encouraging people to use polyc and without the fix it just doesn?t work at all on Mac OS X.
OK, I've just done that. Can you check that it works? In general I don't like replacing public files in place once they've been released but I've made an exception in this case.
I thought I'd checked it on Mac OS X after I made the previous change but it looks like I hadn't.
Regards, David
David,
On 12 May 2014, at 12:08, David Matthews <David.Matthews at prolingua.co.uk> wrote:
On 11/05/2014 17:21, Rob Arthan wrote:
David,
On 11 May 2014, at 12:23, David Matthews <David.Matthews at prolingua.co.uk> wrote:
On 10/05/2014 15:07, Rob Arthan wrote:
The line that sets EXTRALDFLAGS for Mac OS X in polyc.in is missing a double quote character and this causes polyc to raise a syntax error.
I've fixed that in SVN trunk and the fixes-5.2.2 branch. I wonder whether it is worth exporting that and updating the tar.gz file on SourceForge.
I think it might be worth fixing the 5.5.2 release tar.gz file as you have been encouraging people to use polyc and without the fix it just doesn?t work at all on Mac OS X.
OK, I've just done that. Can you check that it works? In general I don't like replacing public files in place once they've been released but I've made an exception in this case.
Nothing seems to have changed - the line that sets EXTRALDFLAGS is still wrong.
Regards,
Rob.
On 13/05/2014 21:27, Rob Arthan wrote:
OK, I've just done that. Can you check that it works? In general I don't like replacing public files in place once they've been released but I've made an exception in this case.
Nothing seems to have changed - the line that sets EXTRALDFLAGS is still wrong.
Sorry. I did something stupid while I was rerunning the export. I still had the existing polyml-5.5.2 directory in my /tmp so mv polyml polyml-5.5.2 put polyml as a sub-directory of the existing directory rather than changing its name!
I have run it again and downloaded to check and it looks as though it now has the fixed version.
Regards, David
On 13 May 2014, at 21:27, Rob Arthan <rda at lemma-one.com> wrote:
David,
On 12 May 2014, at 12:08, David Matthews <David.Matthews at prolingua.co.uk> wrote:
On 11/05/2014 17:21, Rob Arthan wrote:
I think it might be worth fixing the 5.5.2 release tar.gz file as you have been encouraging people to use polyc and without the fix it just doesn?t work at all on Mac OS X.
OK, I've just done that. Can you check that it works? In general I don't like replacing public files in place once they've been released but I've made an exception in this case.
Nothing seems to have changed - the line that sets EXTRALDFLAGS is still wrong.
I tried again and it works now. Perhaps sourceforge had pointed me to a mirror that hadn?t been updated.
Regards,
Rob.
-
David.Matthews＠prolingua.co.uk -
Ramana.Kumar＠cl.cam.ac.uk -
rda＠lemma-one.com