On Wed, Jan 30, 2013 at 07:13:25PM +0100, Peter FELECAN wrote: > "Maciej (Matchek) Bliziński" <[email protected]> writes: > > > While I'm wresting with the memory leak in catalog indexing, I got the > > idea that instead of unpacking and indexing everything locally, we > > could make people's private build hosts download data from our > > buildfarm. What do you think? > > Your proposition has merit but I think that we should keep things > reproducible even without OpenCSW's infrastructure; in an extreme case, > somebody wishing to build in an unconnected bunker (viz. a bank or stuff > like that) can do it.
Sure. I wouldn't think about this data load as of a dependency, but just as an optional speedup. For the last month and a half or so, instead of being productive and playing Call of Duty, I've been rewriting parts of pkgdb and part of that was re-indexing all out packages. I didn't make it run in one smooth go ‒ I will do that after I've finished the main set of changes ‒ because I had to deal with some code issues along the way, but I can tell it took something like 30h to index one catalog (e.g. unstable or dublin). So if you could instead just download a SQL dump file and load it, it would be much quicker. When I'm done, I'm actually thinking of not re-running the indexing on the buildfarm, but instead taking the data from my virtual machine and loading them onto the buildfarm database. Maciej
signature.asc
Description: Digital signature
_______________________________________________ maintainers mailing list [email protected] https://lists.opencsw.org/mailman/listinfo/maintainers .:: This mailing list's archive is public. ::.
