On Tue, Jun 11, 2013 at 12:40:58AM +0200, Jason A. Donenfeld wrote:
> ---
> smtpd/mda.c | 12 +++++++++---
> 1 file changed, 9 insertions(+), 3 deletions(-)
>
> diff --git a/smtpd/mda.c b/smtpd/mda.c
> index 23d288e..24c7442 100644
> --- a/smtpd/mda.c
> +++ b/smtpd/mda.c
> @@ -447,9 +447,15 @@ mda_imsg(struct mproc *p, struct imsg *imsg)
>
> /* update queue entry */
> if (error) {
> - queue_tempfail(e->id, 0, error);
> - snprintf(buf, sizeof buf, "Error (%s)", error);
> - mda_log(e, "TempFail", buf);
> + if (strncmp(error, "BOUNCE:", 7) == 0) {
> + error += 7;
> + queue_permfail(e->id, error);
> + mda_log(e, "PermFail", error);
> + } else {
> + queue_tempfail(e->id, 0, error);
> + snprintf(buf, sizeof buf, "Error (%s)",
> error);
> + mda_log(e, "TempFail", buf);
> + }
> }
> else {
> queue_ok(e->id);
I don't really like that idea:
- error is an informative string the mda will output for a user, we are
not supposed to alter scheduling behaviour based on that;
- the "bounce:" prefix in an error message does not imply a permfail by
any convention/standard I know. Is it ? where did you get that from ?
Can you explain what use-case you're trying to solve ?
We're probably too conservative with mda errors but I would rather see
tempfail/permfail solved by sysexits(3) than by substring of the error
message :-/
--
Gilles Chehade
https://www.poolp.org @poolpOrg
--
You received this email because you are subscribed to mailing list:
[email protected]
To unsubscribe, send mail with subject:
[[email protected]] unregister
