> > Denis, would you like to look into this? It looks like an easy, > localized fix, and your build-repcache patch will benefit from it, too.
Thank you for the suggestion. I'll keep it in mind, but as I already have two outgoing patches, I would rather try not to add one more just now, to keep things focused. Speaking of the possible starvations, I would like to note, that since the current 'build-repcache' implementation uses one SQL transaction per revision, it seems to me that it should behave similarly to concurrent commits, where a rep-cache.db is updated in the post commit processing stage. Regards, Denis Kovalchuk
