Chris Marshall
2014-08-26 12:26:23 UTC
Thanks, Rob!
Chris, https://github.com/mohawk2/pdl-code/ on branch inline-eud
I was slightly surprised to hear the code isn't on the sourceforge repo.
It was easier to push a devel release than to figure out howChris, https://github.com/mohawk2/pdl-code/ on branch inline-eud
I was slightly surprised to hear the code isn't on the sourceforge repo.
to apply your changes to create a branch on the main sf.net
site. If the fork had been on sf.net it would have been easier
to accomplish.
The way the PDL git workflow is operating at the moment isn't very
clear to me. > Are the various branches and their intended purpose
(eg next maint release, PDL3 etc) written down anywhere?
The master is pretty much the main PDL-2.x developmentclear to me. > Are the various branches and their intended purpose
(eg next maint release, PDL3 etc) written down anywhere?
and release area and the idea is that code changes there
should not break the build. I.e., don't commit until tests
pass on at least one platform and you have reason to believe
that the changes will not affect other platforms.
Breaking changes need to be developed in a separate branch.
Craig started the core-cleanup branch removing some of the
cruft from the PDL-2.x core and I started from there to implement
fixes for 64bit index support in the longlong-double-fix branch.
I thought a "quick fix" to the remaining PDL-2.x problems
with 64bit indexing would be the way to go as a good point
to stabilize the PDL-2.x development and start work on the
new PDL3 stuff. Unfortunately, I've been stuck at the debug
stage in longlong-double-fix and only make progress in
fits and starts when I can grab a chunk of time to spin up
all over again.
(It would really help if there were other developers helping
out but apparently most PDL users and developers aren't
having problems with the fact that PDL can't work reliably
with indexes >32bits.)
It has been disappointing to me since it has also been
taking time away from any PDL3 development which is
the real fix to the problems of PDL-2.x and the kludgy
types system.
I've copying this to pdl-porters for other developers to comment
on.
Cheers,
Chris
Cheers,
Ed
Sent: Monday, August 25, 2014 10:30 AM
To: Chris Marshall
Cc: Guggle Worthless
Subject: Re: [Perldl] CHM/PDL-2.007_04.tar.gz released to CPAN
-----Original Message----- From: Chris Marshall
and I was going to fix it there. But I can't find it there, so I guess it's only
on mohawk's git repo.
Ed - if you want to insert the code in (the attached) with.txt into the
END{} block before rmtree() gets called, go ahead. Otherwise I'll attend to
it when it hits the PDL sourceforge repo.
Essentially it's just the same fix as the one to which bulk88 pointed us wrt
Inline::C's TestInlineSetup.pm.
Cheers,
Rob
Ed
Sent: Monday, August 25, 2014 10:30 AM
To: Chris Marshall
Cc: Guggle Worthless
Subject: Re: [Perldl] CHM/PDL-2.007_04.tar.gz released to CPAN
-----Original Message----- From: Chris Marshall
Thanks, Rob. This is supporting some new Inline stuff
that I'm not familiar with (compliments of mohawk).
Duh! - I was thinking inline-with.t was already on PDL's sourceforge repothat I'm not familiar with (compliments of mohawk).
and I was going to fix it there. But I can't find it there, so I guess it's only
on mohawk's git repo.
Ed - if you want to insert the code in (the attached) with.txt into the
END{} block before rmtree() gets called, go ahead. Otherwise I'll attend to
it when it hits the PDL sourceforge repo.
Essentially it's just the same fix as the one to which bulk88 pointed us wrt
Inline::C's TestInlineSetup.pm.
Cheers,
Rob