Everyone,
Here are some ideas I have been floating around in my head that I think might
be useful to do for a possible 1.8 release. If you have things that you
want to address or fix, please bring them up.
My thinking is that I would like to target "small to medium architectural"
issues for this release. I've got some examples below; basically, not the
great Mime Rewrite, but smaller sub-optimal design decisions that have
accumulated over the years. And anything that results in crappy usability
issues.
- Rototill smatch() (argument processing). We have a lot of programs that
want to pass down switches to OTHER programs (the biggest example is
post, which is called by send, whatnow, rcvdist, and probably others).
Some switches take arguments, and some do not. Teaching smatch how
to do things like "build an argument vector for a subcommand" would be
a lot more robust and reduce code. In addition, if we could have a switch
that would let you get all of the switches that a command supports and
their arguments, that would make writing a postproc a lot easier.
See this for background discussion:
http://lists.nongnu.org/archive/html/nmh-workers/2018-01/msg00140.html
- Deal with shell metacharacters when running subcommands or storing
files. We have talked about this, but a common solution and an agreement
on what kind of guarantees we make here would be useful. Background
discussion here:
http://lists.nongnu.org/archive/html/nmh-workers/2018-01/msg00011.html
(My feelings is when we expand things that would run a command
we should simply replace all shell metacharacters when doing
substitutions on the command and commit to that API in the future, but
there was not agreement on that; perhaps just making that the default
and providing a way to NOT do that if you know what you are doing would
be acceptable).
- That thing I owe David a response to from a few months ago (long story,
I document it here as a reminder).
- Decide what to do when we encounter invalid base64 during message display.
- If we can't find a valid db/dbm library, simply don't compile that code
rather than failing configure.
- Write a man page explaining about the network security for inc/post and
friends and how it works in nmh, since it's kind of confusing.
- Question for Eric Gillespie: is the OAUTH2 SASL mechanism the same as
OAUTHBEARER? If so, we should add that support to the code, which would
be only a few lines. If it is different, could you explain how it is
different?
Hmmm ... it looks like it is signifcantly different, at least in some of
the details in that is uses a GS2 header. Dang. Maybe not a huge amount
of code, and we could get it working ...
- All of the bugs on the current bug tracker:
http://savannah.nongnu.org/bugs/?group=nmh
A large number of which have been submitted by Ralph :-) It would be
nice if we could go through and fix them all, or decide if we will never
fix them.
--Ken
--
nmh-workers
https://lists.nongnu.org/mailman/listinfo/nmh-workers