Re: [Nmh-workers] #ifdef UCI

[Valdis . Kletnieks]

On Thu, 05 Jan 2012 11:55:25 EST, Ken Hornstein said:
> In the "garbage collect old crap" vein ...
>
> I've noticed that there are a number of places that are #ifdef UCI.
> They turn on various random things (like, if you have a file called
> .signature in your home directory, it will use it as the signature
> entry in your context).  You would have had to turn this code on by
> hand, and I doubt anyone has used it in a long time.  Anyone care if
> I nuke it?

exmh provides its own .signature handling because it can't guarantee
that the underlying mh package was built with UCI enabled.

Looking at the code real fast...

uip/send.c and uip/whatnowsbr.c:  auto-append .signature file when sending
This is probably Bad and Wrong, because if you're sending a MIME
message (like this one), the appended text will end up outside the
MIME boundaries and thus not displayed by the recipient MUA.  Should
be heaved over the side.

whatnowsbr.c and big chunk of send.c look identical, probably should
be abstracted into a libnmh.so.  Just sayin'.

rmf.c: We loop across a directory, and try to unlink messages one
at a time.  The code skips over files that start with BACKUP_PREFIX,
a ",", or (if MHE is defined a "+").  For UCI, we add "_" and "#" to the
list.  Does current mh-e use + as the deleted flag?

That code needs to be looked at - basically BACKUP_PREFIX
is defaulted to "," anyhow.  The question is if nmh was configured with
some other value, should it continue to skip over ",nnnn" files anyhow?

anno.c: if invoked as 'fanno', it forces -nodate as a parameter.  Not sure
why anno vs fanno mattes here? Anybody know the logic/history here?

conflict.c: Turns on some "group leader" code I don't pretend to understand
at first reading.

pgp0u3NeozgBA.pgp
Description: PGP signature