Re: [Nmh-workers] Questionable code - the bigger picture

[Jon Steinhart]

There's really something bigger going on than just my query about this piece
of questionable code.  A side-effect of a project that I'm working on is that
I'm adding some features to nmh to support that project.  So I'm looking at a
lot of the code.  I'm adding comments to stuff when I figure out what it does,
and would like to clean up things that could use it while I'm at it.  In other
words, I'm trying to incrementally revamp the code.  So I don't just want to
put off all changes for some future grand revamping, because I don't expect
that that will ever happen.

So what's the best way to go about this sort of thing?  One of the biggest
handicaps is the absence of any guidance in the form of requirements or
architecture documentation.  There's no context for making decisions on what
sort of changes are worthwhile.  Can we agree on some way to work this?

Oh, just for discussion, here's another change that I'd like to make.  I need
to make some changes in the way the context/profile stuff is handled.  Without
going into a lot of detail here, I'm adding a -nochangecontext option to many
of the programs so that you can, for example, show +newfolder without changing
the current folder.  Yes, this is amazingly complicated to do.  So aside from
this, the change that I would like to make is to collect all of the context
handling functions into single sbr/context.c file.  It's much easier (at least
for me) to make changes that affect this stuff when it's all in one place.
It's only a few hundred lines of stuff; it's not like it's a huge file.  Any
opinions on this?  I'd personally like to extend this to having all of the
sequence stuff in a single file, all of the folder stuff in a single file,
etc.  Just more maintainable to me.

Jon