Michael Haardt wrote:
> You are right, there is a bug when using :mime. Amazing how it never
> showed up in weeks of sending vacation notices here before, but things
> just don't get tested better by using them longer.
How do you know? If you are sure that anyone that got such a
mal-formated auto-reply would contact you about it, you correspond
with better people then me :-)
> which I also sent to Philip. string_cat() does leave room for a
> terminating NUL byte, but does not terminate the string. In this place
> of the code, I managed to forget terminating the string. I looked at
> the rest, but appearantly I did it right there. Thanks for the bug
> report!
Patch worked like a charm. Thank you very much for resolving this so
quickly. I hope there's still time for this to go into 4.50. Have a
virtual beer on me (or if you're ever in Norway, a real one)...
>>My thinking was that a script /test/ should never write anything to that
>>directory, so it wouldn't hurt just to hard-code it when running the
>>test. I could be wrong though...
>
> If you would give it a vacation directory, it would use it. That's why
> it does not have any right now. ;-)
That's too bad. Having Exim syntax check Sieve scripts is a really nice
feature. As a temporary solution, how about changing line 274 of
filtertest.c to use the return of getenv("SIEVE_TEST_ONCE_DIR") instead
of NULL for vacation_directory. That way it's still possible to test
Sieve filters using vacation.
>>And one last question: In a virtual environment would it be safe to use
>>one sieve_vacation_directory for all users?
>
> No, each user must have its own vacation directory or you may run into
> databases being used for more than one user.
Then isn't that a problem for aliases as well?
Bob
