diff options
| -rw-r--r-- | NEWS | 2 | ||||
| -rw-r--r-- | driver.c | 13 |
2 files changed, 6 insertions, 9 deletions
@@ -8,6 +8,8 @@ fetchmail-4.6.3 (): the choices in various panels. * Added `properties' option for extension scripts. * gcc -Wall cleanup +* Removed ad-hock rewrite of Return-Path introduced in 4.6.1. This was + redundant with what reply_hack already does. There are 250 people on fetchmail-friends and 294 on fetchmail-announce. @@ -543,9 +543,9 @@ int num; /* index of message */ * mailserver's SMTP thought *it* was responsible for final * delivery. * - * Stash away the contents of Return-Path for use in generating - * MAIL FROM later on, then prevent the header from being saved - * with the others. In effect, we strip it off here. + * Stash away the contents of Return-Path (as modified by reply_hack) + * for use in generating MAIL FROM later on, then prevent the header + * from being saved with the others. In effect, we strip it off here. * * If the SMTP server conforms to the standards, and fetchmail gets the * envelope sender from the Return-Path, the new Return-Path should be @@ -555,16 +555,11 @@ int num; /* index of message */ * be passed through as a way of indicating that a message should * not trigger bounces if delivery fails. What we *do* need to do is * make sure we never try to rewrite such a blank Return-Path. We - * handle this with a check for <> in the rewrite logic. - * - * If the Return-Path header has no host part, tack on "@localhost" - * in order to pacify sendmails that want to see an FQDN. + * handle this with a check for <> in the rewrite logic above. */ if (!strncasecmp("Return-Path:", line, 12) && (cp = nxtaddr(line))) { strcpy(return_path, cp); - if (!strchr(return_path, '@')) - strcat(return_path, "@localhost"); if (!ctl->mda) { free(line); continue; |