[Zope-dev] Xron fragility

[Kapil Thangavelu]

A smoke break later, some possible more thoughts and snake oil,

there are two general purpose facilities embodied in xron that i would
like to see made available as zope services, async zope processing and
scheduling.

async--  i'd like to be able to designate TTW that a method execute
outside of the current request/response pair to do some long/heavy
processing.

the design might be easier if this functionality was separated out. a
small dynamic pool of publishing threads to execute async functionality
(hmmm.. why reinvent the wheel, zope threads are already in place... but
what are the consequences of tying up the fixed number of zope's
standard publishing threads on a busy site). and a scheduling thread to
manage scheduled events/scheduling and dispatching to publishing
threads.

the possiblities of such are complicated by the magic number limit of
thread connections to the ZODB. I was wondering how this magic number
came about, is it a performance number dictated by the python
interpretor lock?