On Tue, Jun 10, 2014 at 11:27 AM, Surya Gaddipati
<[email protected]> wrote:
> 'my pull requests view' is a query to mongodb  on pusher field and doesn't 
> involve
> traversing all the builds in memory.

Exactly, we need to be able to able to write various specialized views
(or remote API endpoints, or RSS/Atom feeds, …) which run some kind of
database query without having to load actual hudson.model.Run objects
into memory. No exotic use cases need to be invoked here; just showing
regular build history, test result summaries, etc. demonstrate the
problem. The main workaround we have been using after lazy-loading was
introduced in 1.485 is to just truncate results past, say, 20 or 100
builds counting from the most recent. In a few cases we keep separate
ad-hoc caches of crucial information, like PeepholePermalink does. But
just fixing bugs in the specified behavior of AbstractLazyLoadRunMap
is hard enough, never mind dealing with the flood of problems caused
by core or plugin features unintentionally loading all builds and thus
thrashing the server.

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to