[Pdl-porters] rough longlong-double-fix in git

[Chris Marshall]

All-

I've just pushed the final fixes to the first
rough cut work to finish full 64bit support
for PDL-2.x.

It isn't particularly pretty but I would appreciate
test builds, feedback, or especially debugging
and fixes to the git code.

This is the longlong-double-fix branch in the
sf.net git repo:

http://sourceforge.net/p/pdl/code/ci/longlong-double-fix/tree/

Thanks in advance,
Chris

[Chris Marshall]

More fixes but still a lot of subtest failures. If anyone
can try a build/test on their platform, please report to
the pdl-porters list. You should be able to pull a clean
longlong-double-fix branch from the sf.net git PDL repo
and ( perl Makefile.PL && make && make test )...

One specific problem I'm seeing at this point appears
to arise when a piddle and a perl scalar are combined
(binary op of some sort) and the perl scalar part just
vanishes. An example of this problem is in t/basic.t
and subtest 10 where the xlinvals() calculation seems
to drop the lower bound value on line 247 of Basic.pm

Any thoughts anyone?

--Chris

Post by Chris Marshall
All-
I've just pushed the final fixes to the first
rough cut work to finish full 64bit support
for PDL-2.x.
It isn't particularly pretty but I would appreciate
test builds, feedback, or especially debugging
and fixes to the git code.
This is the longlong-double-fix branch in the
http://sourceforge.net/p/pdl/code/ci/longlong-double-fix/tree/
Thanks in advance,
Chris

[Chris Marshall]

FYI: all my testing is on cygwin64+win7 with badvalues
on, not using NaN for badvalue float/doubles, and not
with per-piddle bad values.

--Chris

Post by Chris Marshall
More fixes but still a lot of subtest failures. If anyone
can try a build/test on their platform, please report to
the pdl-porters list. You should be able to pull a clean
longlong-double-fix branch from the sf.net git PDL repo
and ( perl Makefile.PL && make && make test )...
One specific problem I'm seeing at this point appears
to arise when a piddle and a perl scalar are combined
(binary op of some sort) and the perl scalar part just
vanishes. An example of this problem is in t/basic.t
and subtest 10 where the xlinvals() calculation seems
to drop the lower bound value on line 247 of Basic.pm
Any thoughts anyone?
--Chris

Post by Chris Marshall
All-
I've just pushed the final fixes to the first
rough cut work to finish full 64bit support
for PDL-2.x.
It isn't particularly pretty but I would appreciate
test builds, feedback, or especially debugging
and fixes to the git code.
This is the longlong-double-fix branch in the
http://sourceforge.net/p/pdl/code/ci/longlong-double-fix/tree/
Thanks in advance,
Chris