Re: [Nmh-workers] A question for the nmh historians

[Jon Steinhart]

Jerry wrote:
> > Would it make sense to force sequences to be private if
> > MHCONTEXT is set?
> 
> My gut feel is that nmh programs should follow the -public and
> -private switches and what the mh-sequence(5) manpage says:
> 
>     In general, the commands that create sequences (such as pick
>     and mark) will create public sequences if the folder for which
>     the sequences are being defined is writable by the nmh user.
>     For most commands, this can be overridden by using the switches
>     -public and -private.  But if the folder is read-only, or if
>     the "mh-sequences" profile entry is defined but empty, then
>     private sequences will be created instead.
> 
> But that's just a guess.  I'm also confused about why you say
> the MHCONTEXT machanism wasn't working properly when the other
> user had cur as a public sequence.  All of my folders have cur
> as a public sequence; changing MHCONTEXT hasn't been a problem. (?)

A primary reason to use MHCONTEXT is so that you can have mh programs
operate independently.  For example if in one shell I'm reading mail
in the usual way, i.e., show, next, next, next, ..., and in another
shell I can do

        MHCONTEXT=/tmp/anothercontext
        export MHCONTEXT

        show first +anotherfolder

The above shouldn't affect the usual mail reading because it uses a
different context.  The problem is that it doesn't completely work
unless the cur sequence is private.  That's because a public cur
sequence is stored in the .mh_profile, not in the context file.  So
it's impossible to completely break the context interaction.  The
folder will be independent but the message won't.

This seems like a bug, or at least something poorly thought out to
me.

Jon