On Tue, 2005-03-15 at 22:34, Philip Hazel wrote:
> I have now got round to reviewing it, and checking out the
> circumstances. It turns out that it is in fact a bug in a fix for a
> previous problem (ChangeLog 4.11/85). It also turns out that there is a
> neater patch than the one you posted (it doesn't produce any debugging
> output, however). I'll copy it below (for 4.50).
>
> Thanks for the diagnosis. It made my job a lot easier.
Sorry for the late reply. This didn't dawn on me until I was
daydreaming on the way home.
Your patch is very similar to my original attempt, which I
posted here:
http://www.exim.org/mail-archives/exim-users/Week-of-Mon-20050221/msg00334.html
That did fix the original problem, but introduced new bugs.
I didn't elaborate at the time - but it caused exim to
drop messages. I didn't look into why too closely, because
I thought the patch was suspect from the start -
previously_transported mungs the linked lists that
do_local_deliveries is traversing at the time. So in the
end I opted for a less invasive solution that just
skipped the unwanted deliveries.
The reason for this long spiel is I have not tested your
patch, but it does look like it would exhibit the same
faults as my original one. I think my test script did
demonstrate the fault - if you go looking at what isn't
delivered.
