jjc@UNIX.CIS.PITT.EDU ("Jeffrey J. Carpenter") (10/21/90)
Hi, I am running ATK at patch level 7 (most recent). I am using the white pages but not amds. The problem I am having is that when sending a message with messages, et. al. the address xyz+ is accepatble, but addresses in the format xyz+aaa result in validation failure. Any ideas? Thanks, jeff Jeff Carpenter University of Pittsburgh, Computing and Information Services 600 Epsilon Drive, Pittsburgh, Pennsylvania 15238 jjc+@unix.cis.pitt.edu, jjc@pittvms.bitnet, +1 412 624 6424, FAX +1 412 624 6436 [An Andrew ToolKit view (a raster image) was included here, but could not be displayed.]
nsb@THUMPER.BELLCORE.COM (Nathaniel Borenstein) (10/22/90)
If you aren't running AMDS, I don't think that xyz+ should validate either, because the delivery system is going to end up rejecting it anyway. (Unless, of course, it turns "xyz+" into simply "xyz" in which case it is harmless.) If you're not running AMDS, there is no way that the xyz+abc form will be deliverable, so it makes sense for it not to validate. Thus the bug, if any, is that xyz+ does validate. And, as I said, if it removes the "+" in the validation process then I don't see where there is any real bug...
Craig_Everhart@TRANSARC.COM (10/23/90)
As Nathaniel says, without AMDS the forms with plus signs mean nothing. Here's what I believe is happening. Since your domain (unix.cis.pitt.edu) doesn't run AMDS (and since AMS knows it), the ``+'' character looks to the heuristic name matcher just like other noise/punctuation characters. Thus, when you compose mail to jjc+@unix.cis.pitt.edu, the validation routines turn it into jjc@unix.cis.pitt.edu before writing it into mail headers, so that other sites can reply to it. In particular, your From: address has no plus-sign in it. Or have you hacked your local sendmail to notice the ``+'' character as a UID delimiter? Even if so, there's no simple way to tell AMS that your site uses the ``+'' delimiter rule other than by turning on AMDS; you can, if you like, disable WP validation entirely, but that doesn't sound like what you want to do. (Yes, it would be nice to have an option to tell AMS to use the ``+'' convention independent of AMDS.) We're guessing at your local config and at what you want to do; feel free to fill in the gaps. But in any case the ``+'' is a delivery-system configuration, not something that comes bundled with WP. Craig
jjc@UNIX.CIS.PITT.EDU ("Jeffrey J. Carpenter") (10/23/90)
Excerpts from comp.info-andrew: 22-Oct-90 Re: Problem with ams Craig_Everhart@transarc. (1171+0) > Or have you hacked your local sendmail to notice the ``+'' character as > a UID delimiter? Even if so, there's no simple way to tell AMS that > your site uses the ``+'' delimiter rule other than by turning on AMDS; > you can, if you like, disable WP validation entirely, but that doesn't > sound like what you want to do. (Yes, it would be nice to have an > option to tell AMS to use the ``+'' convention independent of AMDS.) > We're guessing at your local config and at what you want to do; feel > free to fill in the gaps. But in any case the ``+'' is a > delivery-system configuration, not something that comes bundled with WP. > Craig We did hack up sendmail. At one point this did work, but I am not sure what comination of options and versions it worked under. We were able to send mail to something like jjc+xxx@unix, and it would take it. Also, when it resolved names on out system, it would add the + to the end (if I typed in jjc@unix, it would become jjc+@unix.cis.pitt.edu after validation). Eventually, we will be running AMDS, but not yet. Our problem is that we can not get rid of sendmail yet. The reason we want to do this is because we want to run a bulletin board facility through AMS using flames. Everything works but this one part. jeff
nsb@THUMPER.BELLCORE.COM (Nathaniel Borenstein) (10/23/90)
Sigh... sounds like what you need is to separate the current
AMS_NonAMSDelivery AndrewSetup option into two options, one that says
you're not running AMS Delivery, and one that says you're allowing "+"
validation anyway, or something like that.
I suspect this wouldn't be too hard to do, because the only two things
I'm sure you'd need to change are 1) the test on line 197 of
overhead/mail/lib/locname.c, and 2) the test on line 1690 of
ams/libs/ms/mswp.c. In both cases, these test to see if you're running
AMS Delivery when you might prefer to have them testing for the new
configuration value you would add.
To actually add the new AndrewSetup option, you only need to change two
files, overhead/mail/lib/mailconf.{c,h}. These changes should be pretty
obvious.
If you make these changes, you should probably post them where Craig & I
can look them over, and then submit them to the ITC for consideration
for future patches. -- NathanielCraig_Everhart@TRANSARC.COM (10/23/90)
It's somewhat harder than that (locname.c, mswp.c, mailconf.{c,h});
there are about 30 places in overhead/mail and ams/* that conditionally
use ``+'' as part of an address. I've wondered about doing something
comparable, since it would help support cs.cmu.edu as well. To be
useful in that context, though, you'd have to make it a per-cell option:
it would have to be part of the structure that
overhead/mail/lib/ckamsdel.c manages.
In any case, please don't just add it as an AndrewSetup option, which
could apply to only one site at a time.
Maybe I should work on this. Sigh, indeed.
Craig