On Tue, Feb 15, 2005 at 04:30:29AM +0100, Bob Johannessen wrote:
> /* Define this for development of the vacation Sieve extension. */
> /* The code is not yet finished. */
> #define VACATION
>
> It tells me the code is for Sieve vacation /development/ and
> not yet finished, but then proceeds to define VACATION anyway.
Oops. That certainly does not look good.
> So, I guess my first few questions would be; is the current
> Sieve vacation code safe to use? What, if anything, is still
> not finished? Are anyone working on this, or should I start
> looking into it myself? And if so, has anyone got any hints
> on where to start?
The code is not well tested (read: it works for me and some other people
I know). Apart from that, it is finished, that's why it is enabled now.
I should have sent Philip a patch to remove the comment. Philip, are
you reading this?
> Second issue: Testing Sieve scripts with exim -bf ends up
> calling sieve_interpret() (the Sieve script entry point)
> with uschar *vacation_directory set to NULL (.../src/filtertest.c,
> line 274 in the latest snapshot). This results in an error
> of "vacation disabled" in parse_start() when "require"-ing
> "vacation" (.../src/sieve.c, line 2665).
That's correct.
> Is there any reason not to hardcode *vacation_directory to
> for example "/tmp/sieve-vacation" when testing a script?
Vacation without a vacation_directory looks mostly useless to me, but
I am interested to hear different opinions. I could probably add some
code specifically for testing that does not require a vacation directory
and does not create or access files, so you could feed it filter files
that use the vacation extension.
Michael
