Re: [Monotone-devel] Re: Keyword substitution?

[Bruce Stephens]

"Todd A. Jacobs" <address@hidden> writes:

> On Tue, Aug 02, 2005 at 06:55:42PM +0100, Bruce Stephens wrote:

[...]

>>I suspect it's just as likely that you'd have some file (a script or
>>bit of config or something) that came from some revision but had
>>(possibly) been changed a little.  You'd want to know where it came
>>from to know how it had been changed, and perhaps how it had been
>>revised in the repository.  And in that case searching for its hash
>>would likely be a waste of time, because it's probably been changed.
>>Having a keyword of some kind might help, in that case.
>
> Yes, that's exactly it. :) And of course, in a widely-distributed
> monotone group, some people may or may not be tracking various
> branches so being able to tell which repository a particular file
> came from would be helpful, too.

OK, that sounds like a use case: you want to be able to export a
revision of files, and later (perhaps after they've been changed),
discover where they came from.  

(You aren't alone.  At work we have occasion to do that kind of thing.
Most of our code is binary, but there's a fair amount of scripting,
and occasionally we've suggested customers patch the scripts and maybe
just "oh, try this one" from a later version.  I don't think we've
ever lost track to the extent that we need to look at the CVS
keywords, but I can imagine it happening.)

That sounds like it could be done as a separate operation largely
outside monotone: you check out a tree, and then apply some process to
add that information to each file, or perform substitutions in each
file.  Basically what monotone does when it builds itself, I guess.
Would that do?  So a script or two in contrib/ might be just as good
as monotone itself performing the substitution?

[...]