but this makes me think of something else... we currently have
"automate stdio", that lets a persistent monotone process keep db
startup and cache contents across individual automate commands..
what about a "monotone interactive" (or just plain monotone with no
command?) that drops the user at a monotone> prompt, takes lines of
input just as if they were in argv, and wraps readline and
tab-completion (for subcommands too) around it based on the current
expansion mechanism.
So a typical workflow might go something like:
% monotone
monotone> status
...
monotone> list unknown
...
monotone> add --unknown
...
monotone> commit
...
monotone> pull
...
monotone> merge
...
monotone> sync
...
monotone> update
...
monotone> ^D
Many of these ops would tend to hit the same data in the db each time,
and it might make things a little snappier...
It's (perhaps sadly) not something that is easily provided by a
wrapper program and/or shell completion, because the whole point is to
keep the db open across operations.