RE: new cvs committer, some ideas

Please update the exmh.CHANGES file with a summary of your changes.  It
looks like that was last updated in '06

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Alexander Zangerl
Sent: Wednesday, June 18, 2008 6:23 AM
To: [email protected]
Subject: new cvs committer, some ideas

i'm the debian maintainer of the exmh package and over time we've
accumulated some differences between the (static) upstream and 
the debian version.

in a Sneak Move i convinced Brent Welch last week to add my sourceforge 
account to the list of exmh developers. as az143 i've since committed
most of the bug fixes and enhancements from debian back to the exmh cvs.
(the list of mods follows at the bottom.)

however, there's a few items i'd like to float past you fellow exmh
workers
before i commit them.

* installation paths
debian requires more strict adherence to the filesystem hierarchy
standard,
which means for exmh that bitmaps, helper scripts, documentation and
tcl code should not all end up in /usr/lib/exmh.

the debian version uses /usr/share/exmh/bitmaps for guess what,
/usr/share/exmh/ for scripts, 
/usr/share/doc/exmh for documentation,
/usr/lib/exmh for tcl code and /usr/bin/exmh for the executable.

the code only needed slight modifications to cleanly use
$install(dir,xxx) 
and $exmh(xxx) to access these separate areas; it is thus easy to set
these vars all to /usr/lib/exmh in the installer to get the original
behaviour.

any objections to this kind of split? if i hear no contrary voices
i'll commit that change to the cvs in a week or two.

* explicit paths in #!
most scripts in the exmh package use unqualified interpreter names
for wish (but not for /bin/sh, for whatever reason). personally i find
that
a bad idea, and would very much suggest using /usr/bin/wish 
and /usr/bin/expect explicitely. any comments on that?

regards
az

(list if debian-diffs checked in recently:
some manpage fixes to silence man warnings
the initial no-mh-yet greeting mentions install-mh with path
spamfilter training for ham now also allows passing of messages
 via stdin; before only the spam trainer would honor $bogo(stdin)
the glimpse -W option was promoted to a proper global preference
 instead of being settable only for a single search
my fix for the stale scan cache problem was committed
the unclickable links for inline-displayed html were fixed
support for delegating passphrase handling to gnupg's gpg-agent was
added
marion hakanson's recent patch for the internal mime decoder was
committed
exmh now supports recode as an alternative to mimencode (from metamail) 
 for qp/base64-en/decoding. the sequence is: use mimencode or recode or
 the internal tcl decoder.)


-- 
+ Alexander Zangerl + DSA 42BD645D + (RSA 5B586291)
Kindermund: Ein Pfirsich ist wie ein Apfel mit Teppich drauf.