Christian Soeller wrote: > Many good points have been made in this thread. A couple > observations/questions from my side: > > 1) competition is good. These days I use IDL, NumPy, PDL and others (in > sequence of frequency) at various times. Basically no one tool is good > for every purpose. > > 2) Do we need to handle backwards compatibility, i.e. do other package > maintainers expect 'require => ['PDL']' to work in certain ways and do > we need to honour this?
I definitely think we need to handle/manage backward compatibility going forwards. Like it or not, people who really use PDL won't be happy if all their code breaks suddenly. However, with transition planning, maybe backward compatibility modes, etc we should be able to make needed changes. Some of this would be simpler to track and sort out if our tests were a bit more complete in their coverage. --Chris > Christian > > On 3/11/2009, at 10:09 AM, David Mertens wrote: > >> On Mon, Nov 2, 2009 at 9:54 AM, Doug Hunt <[email protected] >> <mailto:[email protected]>> wrote: >> >> Hi all: I vote we keep PDL::Slatec out of the 'core' PDL. I'm trying to >> get rid of dependencies on this module in my code just to make builds >> easier on my systems. >> >> >> There's bound to be lots of discussion about what should and should >> not go into the core. I think that absolutely no external >> dependencies should go into the core, so Slatec would be out. The >> goal of the core would be to have a module that any Perl module author >> could use, knowing it would install on anybody's machine without any >> hiccups. _______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
