----- Original Message -----
From: "Chris Marshall" <[EMAIL PROTECTED]>
.
.
>
> Thanks for the heads up on the modifications,
> Rob. I think I'll let 03 run for awhile...at least
> until someone fixes the ~ expansion problem.
> :-)
In the meantime, is it ok to add changes to the CVS ?
I have this recurring problem with dmake (only), where, unless PDL::IO::GD
is built, the PDL::IO::Dicom Makefile's postamble section gets stuff written
into it that breaks the build. I don't know where that "stuff" comes from -
if PDL::IO::GD gets built, then the Dicom Makefile's postamble section is
empty, which is as it should be. There's a simple fix - just write a sub
MY::postamble in the Dicom Makefile.PL:
sub MY::postamble {
return "";
}
And that's probably what I'll do if I can't get to the root cause of the
problem. The main trouble is that PDL takes so long to build ... it takes
about 30 minutes to test a change. ie it takes around 30 minutes to run
'dmake clean', and then re-run 'perl Makefile.PL' followed by 'dmake'.
Multiply that time by the number of different builds one might want to test
(various permutations of 5.8, 5.10, Win32, nmake, dmake, linux) and it
becomes a very slow process indeed.
There would be no-one happier than I if PDL were to be trimmed back to a
very minimal core. Not only would that remove many of the problems that we
spend our time dealing with, but it would also enable testing to be carried
out much more quickly.
Cheers,
Rob
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
