http://issues.apache.org/SpamAssassin/show_bug.cgi?id=5269
[EMAIL PROTECTED] changed:
What |Removed |Added
----------------------------------------------------------------------------
Severity|major |blocker
------- Additional Comments From [EMAIL PROTECTED] 2007-01-03 02:31 -------
'First, we can fix the whole issue through the use of a while statement (or an
if
statement at a minimum), so I don't think we need to go redesigning the idea of
what an update revision is, nor do we need to separate 3.1 and 3.2 updates in
different channels, etc.'
If we use a while statement, that assumes
(a) there will never be a race condition between backend update generation
scripts for 3.1.x, 3.2.x (and 3.3.x, 3.4.x et al)
(b) all those scripts write their output to the same filesystem, one that is
race-immune (like a POSIX fs), and to the same directory path -- ie. not synced
into place
So we're locking down update generation, basically, to always happen on the same
machine.
'This is a relatively simple problem that happens
infrequently, so let's keep the solution simple as well.'
Well, if you ask me, race conditions are not simple. Things would look less
simple when you have to add race-avoidance locking to that while loop.
Anyway, go ahead and take a shot at it. The 3.2.x updates won't work until this
happens, so marking as a blocker. FWIW, the code worked fine, apart from the
warning... ;)
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.
