On a fresh install (1.1.4 using pip), successfully running on celery and RabbitMQ, when viewing an repository with >10k commits a single view to the changelog page causes ~300Mb memory usage in paster. Each view of a subsequent page causes another similar amount of memory to be consumed by the same paster process, and never gets released. After page 6, the process is holding on to over 2Gb of memory. Is this expected behavior?
I should also point out that statistics are disabled (thoguh I don't think the changelog page is affected by this), and noone else is browsing this instance during this repeatable test.
Thanks for a great product,