List · Roadmap
STR #18
| Status: | 5 - New |
| Priority: | 3 - Moderate, e.g. unable to compile the software |
| Scope: | 3 - Applies to all machines and operating systems |
| Subsystem: | Newsd Server |
| Summary: | X-X-Loop handling seems a bit broken |
| Version: | 1.44rc4 |
| Created By: | erco |
| Assigned To: | erco |
| Fix Version: | 1.44rc5 |
| Update Notification: | |
Trouble Report Files:
Post File
No files
Trouble Report Dialog:
Post Text
erco 02:05 May 05, 2005 | Original poster: mdorman (see newsd.general of same title)
"Around line 257 of newsd.C, newsd checks for the presence of X-Newsd-Loop or X-Loop headers, and if it finds them, the mail is considered to be part of a mailing loop, and is thus not delivered.
"My understanding of the intent of X-Loop: is that you put some value that is unique to *your* system (something like newsd@news.example.com") in it, and then you look for that token---and if that token that only you would have generated appears, you know you're on the second go-round.
"As it is, since newsd doesn't inject any such header itself, it will only catch loops if someone else happens to inject such a header, so it's pretty ineffectual. But, because it doesn't look for a value that it would have generated, it *also* keeps you from gatewaying any mailing list that uses a mailing-list processor or mail transfer agent that injects that header itself.
"In my local copy I've just diked this out entirely. I would suggest that this be done in the short term, and in the long term, proper x-loop handling be implemented. I'm willing to look into doing this...but later, unfortunately."
|
erco 02:13 May 05, 2005 |
Comments on the above: redux of newsgroup thread to date:
Will investigate injecting X-Newsd-Loop into the headers on outbound mail gatewaying, and make some abusive aliases that redirect back to the newsd mail alias to see if it gets properly shunted to .deadletters.
Also, regarding "deliver to Errors-To:" comment in CHANGES file: this is not actually implemented, and either should be (after safely making sure loops can't occur), or the comment should be striken and left unimplemented. |
|