On Fri, 2003-11-28 at 17:43, Chris Cheney wrote: > On Wed, Nov 26, 2003 at 06:25:46AM +0000, Henning Makholm wrote: > > In the former case, would better algorithms help, or is it > > intrinsically a hard problem? > > Things as large as KDE will likely never be able to migrate to testing > on its own. This is due to the number of packages KDE depends on and the > number of packages that depend on KDE. Since there is such a large > number the likelyhood of at least one of them having a RC bug is > extremely high, which in turn blocks an automatic migration. (At least > AIUI) I wonder how possible it would be to reduce the dependency complexities in a project as large as KDE. Where I used to work, that was something that started to bite as at one point - ~500K lines of code, and we did a little work to separate out the lowest "lib" layer, but always had deadlines to work to, and therefore had difficulty getting management approval for restructuring work. KDE is open source/ community though. every layer can only depend on layers below libs layer is lowest rules for threading code got interesting (locking dependencies/ requirements I think) - I could track this down if someone needs 'em ?more? cheers zen -- Phone: +61 (0)412 166 990 Homepage: http://homepages.ihug.com.au/~zenaan/ PGP Key: http://homepages.ihug.com.au/~zenaan/zen.asc Please respect this email's confidentiality as sensibly warranted.
Attachment:
signature.asc
Description: This is a digitally signed message part