Re: Conflict between exmh cache and scan? (fwd)

On Fri, 02 Jan 2004 18:36:37 PST, Brent Welch <[email protected]>  said:

> Any thoughts on this thread?  The issue is the pattern matching
> that picks out the message number from a scan line.  It is currently
> anchored to the beginning of the line, (not counting white space),
> but this user seems to think that is a behavior change.  Any risks
> with not anchoring the pattern?

Yes.  It's a major stability issue.

> |  > for any email whose scan listing contained an "N" (i.e. messages 29, 30 
> |  > above).  The "offending" line in my scan form file was
> |  > 
> |  > %<{status} %|N%>\

We have to make the assumption that the output of 'scan' is at least somewhat
reasonable, and we really need to just put our foot down and say "the first field
has to be the message ID".  Consider if the offending format item had been this instead:

%6(size)

Yes, output the size as the first field (or any field before the msg number, if
we make the proposed change). Now use that in a folder with about 2,000 or so
messages that vary between 600 and 4K bytes in size.  Have fun guessing what
message will show up when you click on it. :) (Yes, I tried that once, when I
was young and foolish and thought that exmh was using the line number in the
text widget to find things.  I was quickly disabused of that heretical thought
;)

Similar arguments apply to scan formats that output time/date info - particularly
fun if you have one that does 'time' if it's less than 24 hours old, 'day NNmon'
("Fri  1Jan" for example) if it's under a year old, and '05 Dec 02' for stuff over
a year old).  And most of the other scan formats can be abused to produce a
numeric as well...

OK, we *could* say "no numerics before the message id" - that works fine till
somebody puts '%friendly(from)' as the first field and then gets spammed by the
crew that sends with a From: that matches <2[0-9]{5}@' (yes, there actually is
such a spam crew - the From: is the spam run number, which is also placed into
the Subject: line as well).  It's a lot easier to just declare the message
number has to be first.  

The user wants another format when using MH in line mode, that's fine - he's
welcome to fill in the 'Custom Criterion' field on the Scan Listing Preferences
and put a '-form something.sane' for EXMH. ;)

If somebody wants to do something *useful* in this area, have it Do Something
Reasonable when a %(msg) format overflows - for example, here's what happens
with %4(msg) on my Bugtraq folder:

% scan -form scan/timely.1 9997-10003
9997       17 Sep 02    4043 Cody Hatch         Using LaBrea to slow spam <<---
9998       16 Sep 02    2946 Mario van Velzen   Analysis of Modap worm <<Greeti
9999       13 Sep 02    3055 "Steven M. Bellov  Re: Bypassing SMTP Content Prot
?000       17 Sep 02    3003 Nate Lawson        Re: Password Security Policy Qu
?001       18 Sep 02    6084 Ajai Khattri       Re: Linux Slapper Worm <<Not se
?002 3     18 Sep 02    6573 Thor Larholm       Mozilla vulnerabilities, an upd
?003       17 Sep 02    9714 [email protected]  The Art of Unspoofing <<I found 

something of the form 'if [regexp {^( *)?([0-9]+) $line .... { signal_error'
would do it.  Probably need to quote that ? if its magic.  You get the idea.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)
Comment: Exmh version 2.5 07/13/2001

iD8DBQE/9jVacC3lWbTT17ARAgqgAKDeR5jl+mtQt+Hc8sTglFk41iQDJQCeI25m
imw+rj/dFNx625DsgZSW+3s=
=CdcP
-----END PGP SIGNATURE-----