[Peter Samuelson] > Yeah, well ... the --la-file option itself is a poor interface. The > very existence of a ".la file" should be a libtool implementation > detail that libraries and applications never need to know about. They > should be asking "what do I add to my link line to link to neon", not > "what is the filename for the internal libtool metadata for libneon".
Like I said, I don't know what's Correct. Clearly you've got opinions about how things would be in an ideal world. > That said, since 'neon-config --la-file' is a preexisting interface, it > should have been handled better. In practice the question it is > answering is equivalent to "what do I add to my link line", so it > should answer accordingly, meaning it should give the same output as > 'neon-config --libs'. Or, why not simply ship the .la file as libneon26.la, and have neon-config refer to that? There's absolutely no reason that I can think of why this can't all be done in a way that allows multiple versions of neon to coexist. > As for the question of whether the .la file should have been deleted - > I still think it should be deleted. libtool functions fine without a > .la file, the need for this file was artificial created by neon-config > having an option to expose the .la filename, which AFAICT applications > never actually needed. Unfortunately, this seems to break the build of the very software you originally wished to make this change for -- Subversion. Subversion uses 'neon-config --la-file' to find neon.
signature.asc
Description: OpenPGP digital signature
