Re: [Monotone-devel] mtn copy anyone?

[Brian May]

>>>>> "Nathaniel" == Nathaniel Smith <address@hidden> writes:

    Nathaniel> It could just be equivalent to "cp foo bar; mtn add
    Nathaniel> bar".

Make that cp foo bar && mtn add bar

We don't want the add to proceed if the copy failed.

If there was such a command, I would use it (I am not arguing there
should be such a command though; it would also be easy to create a
shell alias)

    Nathaniel>  It could somehow record something special in the
    Nathaniel> history.  This special thing could be nothing but a bit
    Nathaniel> of metadata that was exactly equivalent to an add for
    Nathaniel> history tracking purposes, but that, say, 'log' and
    Nathaniel> 'annotate' did something smarter with.  Or it could

Automatically add to the log message for the next commit "This file
derived from file /path/to/file <fileid> in revision <revisionid>"?

    Nathaniel> actually affect the history model -- then there are
    Nathaniel> questions like how it works through merging.  This is
    Nathaniel> unclear too; there are a few alternative, contradictory
    Nathaniel> "natural" semantics for copy WRT merge.  (There's a
    Nathaniel> little about this at
    Nathaniel> http://www.venge.net/mtn-wiki/HistoryBlueSky.)

I would prefer we stayed out of this mess. Don't explicitly track
history, but maybe record some sort of pointer indicating where the
file came from, if required.

Or maybe track history, but only for some commands like annotate, and
definitely not merging. This might be confusing though.
-- 
Brian May <address@hidden>