Re: [Monotone-devel] Keyword substitution?

[Todd A. Jacobs]

On Fri, Jul 29, 2005 at 05:43:02PM -0700, Nathaniel Smith wrote:

> This seems like something completely outside monotone's scope; if
> you're inserting project-specific data in a project-specific way,

I agree with you that manually inserting data is pointless, although I 
can see some value in having a keyword for the current tag, as the 
previous poster suggested, since keywords wold be canonical and wouldn't 
hork the hash.

However, I think this all might be addressing the problem backwards. 
What if you look at it like this?

Say you have revision foo of file bar, only you don't *know* what 
revision it is; more importantly, it's not the head revision.

How do you figure out what revision it is within a given repository?

If it's *your* repository, and your file, maybe there could be a 
function to search for a matching hash, so that you could determine that 
file bar is actually revision foo from your repository. *That* would 
certainly be useful.

It would, however, also require you to have access to a repository with 
full history and an exact copy of that specific ancestor, which is (as I 
understand it) not an absolute guarantee in a widely-distributed 
monotone network.

And of course, without keywords, you can't identify the file at all 
unless you are tracking the source yourself, so monotone-managed files 
such as FAQ files, web pages, and others would be impossible to identify 
without manual modifications. :)

--
Re-Interpreting Historic Miracles with SED #141: %s/water/wine/g