Re: Performance issues with Cons
[email protected] Mon, 7 Oct 2002 10:44:09 -0400
| Newsgroups | gmane.comp.gnu.cons.general |
|---|---|
| Message-ID | <[email protected]> |
> But you still have to examine each include file to see if it's changed, > all the way down the tree. You can avoid checksumming it if the date > hasn't changed, but even that can be problematic in an NFS environment. I was looking at scons on the weekend, and they have a flag that tells scons to cache the dependancy information - one flag tells it to rebuild the cached dependancy info, and another tells it to just used the cached versions... I think that's the kind of thing I'd like - then devs who know they haven't changed a header since their last build can just used the cached dependancy data... Anyone have any idea how hard that would be to implement in cons? > Cons can avoid re-checksumming the file if the timestamp matches; see > "Using build signatures to decide when to rebuild files" in the cons > docs. Try using > SourceSignature '*' => 'stored-content'; > and see if that helps. (Requires a recent cons version.) I'll give that a try - I doubt it'll save enough time, though... > You are using one of the native-code MD5 implementations, right? The > perl-based one is dog-slow. I'm pretty sure I'm using a native-code implementation - it's the Digest::MD5 module off CPAN - it certainly generates and installs an MD5.so - so I'm assume it's native-code... Thanks for the information... Warren _______________________________________________ [email protected] http://mail.gnu.org/mailman/listinfo/cons-discuss Cons URL: http://www.dsmit.com/cons/