Re: BTS is too slow? (was Re: enough already: xaw-wrappers is fixed)
In article <Pine.LNX.3.96.990626013056.30450L-100000@tweet> you wrote:
> This reminds me of the problem I have with the BTS, one that I've
> neglected to bring up for discussion until now. The problem I see is
> simple: the BTS is too slow.
We built a derivative of the BTS for internal use in my team at HP, and one
of the things we did was to eliminate all of the cron'ed batch processing,
making each arriving message process more or less immediately. This, combined
with the use of server-side includes and summary building on-the-fly using
cgi-bin scripts, left us with near-instant response to updates via email.
This is *amazingly* handy when you're cranking along working through bugs.
I wasn't real happy with some of the other cosmetic changes that got made
along the way (it doesn't look much like the Debian BTS any more), but it
sure is fast...
There are probably some scaling issues with some of the work we did, but I
suspect a hybrid approach could really speed up the BTS. I know that the
person who did the work for me provided a copy of his code at least to netgod,
who expressed an interest in it, but I don't know if anything ever happened
as a result.
I don't really have time to add the BTS to the list of things I'm hacking on,
but if anyone is seriously interested in looking at what we did, I'll be happy
to make the source available.