On Wed, Mar 25, 2009 at 3:20 AM, Steve M. Robbins <st...@sumost.ca> wrote: > Is the NEW queue going to get processed any time soon? There are 215 > packages waiting [1] about half of which have been there 3 or more > weeks.
It might be worthwhile reflecting upon what purpose the queue has. In a simple model, if the inflow rate equals the outflow rate, the queue in between can be arbitrarily long without serving any purpose or function but causing a delay (which is bad!). But, in reality, at least a part of the queue has a buffering function (since the ftp masters cannot work exactly in sync with incoming packages). Another function is as "storage place" during active work on evaluating packages. I guess the former function, on average, requires most of the queue length. It is probably good to have some standard for what is a reasonable length for this "buffering" and define a threshold over which the outflow rate should be higher than the inflow rate. In order to get the outflow rate higher than the inflow rate it is important that the work on the queue is fun and motivating. If the queue is long, people get irritated which doesn't make the important work of the ftp masters more rewarding. So, this is a bad situation. I guess one of the conclusions to be drawn is that it is extremely important to have a sufficient number of ftp masters. The project should prioritize the work on finding more ftp masters. I also think it is questionable to have strict rules for the ordering of processing, since having some degree of flexibility both can adapt the work somewhat to the ftp masters, so that work becomes easier and more fun rather than the reverse, and, make it possible to prioritize in a way that benefits the project as a whole. If queue length can be kept down this shouldn't be a problem. We must remember that having a queue length over a certain threshold serves no purpose at all! -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org