Re: [Gnu-arch-users] Re: [BUG] Re: False "binary" files

[Robert Collins]

On Sun, 2005-02-13 at 20:10 -0500, Stefan Monnier wrote:
> 
> Nb 2 is clearly the most flexible option.
> It could even be used internally to replace the .arch-ids subdirectory
> with
> a .arch-ids file with a specialized merge/diff algorithm (which would
> mirror
> the order-agnostic diff/merge currently used for directories).

No, it can't. Replacing .arch-ids the directory with a file in your
working tree requires changing the tree format: otherwise local
inventories will fail - and this is before diff or merge are used. I've
started writing up some notes on how changing the project tree format
could/should work: http://wiki.gnuarch.org/moin.cgi/ProjectTreeFormat.

If you have a newer tree format, and the same changeset format, then the
changeset still needs to represent changes to external file ids as
patches to .../.arch-ids/*id. (for older clients).

To hook in a 'smart' changeset with the sort of diff hooks you are
describing will require a new changeset format identifier anyway (and a
new archive format identifier as clients of the current formats don't
look for changeset format identifiers). If we are putting that in place,
why not represent id changes directly as data, rather than as punned
files? (thus eliminating this particular use of custom-diff/merge
facility - and still being compatible with
http://wiki.gnuarch.org/moin.cgi/ProjectTreeFormat1 trees.

Cheers,
Rob

signature.asc
Description: This is a digitally signed message part