Todd Lyons wrote, on 17/02/14 17:33:
> On Mon, Feb 17, 2014 at 8:05 AM, Wolfgang Breyha <wbreyha@???> wrote:
>>> Still searching for the source of it. It's not the DKIM code itself and using
>>> "-oi" fixes it instantly.
>> I think I found the reason and a possible fix:
>> http://bugs.exim.org/show_bug.cgi?id=1444
>
> Can you explain a little bit about the logic for this fix? You're
> detecting a \r and then setting a character count (assuming it's a
> buffer to drain), but I don't quite grasp _why_ this is necessary.
No, I'm not setting a character count. I'm setting the state ch_state=2 as
done in all other states if a "\r" is found.
Here are my notes playing through the states and counts for the testinput.
That made it pretty obvious for me:
test\r\n\r\n
ch_st==1 "t"
ch_st=0 -> ll++ == 1; fput "t"
->
ch_st==0 "e" ->ll++ == 2; fput "e"
->
ch_st==0 "s" ->ll++ == 3; fput "s"
->
ch_st==0 "t" ->ll++ == 4; fput "t"
->
ch_st==0 "\r" -> ch_st = 2
->
ch_st==2 "\n" -> bodylc++ == 1; ch_st = 1; ll=-1; ll++; fput "\n"
->
ch_st==1 "\r"
Greetings, Wolfgang
--
Wolfgang Breyha <wbreyha@???> |
http://www.blafasel.at/
Vienna University Computer Center | Austria
