[darcs-users] Re: 'darcs annotate' output seems rather cryptic

[Nils Decker]

David Roundy <droundy at abridgegame.org> wrote:
> But since darcs changes have long
> names, it is very hard to read.

I have used cvs and svn before switching to darcs. Cvs and svn are nice,
because there is a concept of 'timeline' to the different revisions.
This timeline is easy on my mind, because i can intitivly and easily
find and name the different states in history.

I know that there can't be such a timeline for a distributed darcs
colletion of repos. But every single darcs repo has a straigt history of
applied patches. Those patches could be numbered.

There are some commands that can change the order of the patches.
Unpull, optimise and repair are candidates for history disrupting
commands.

Why bother? The number of the patches can be used as short identifiers
for patches. I saw a suggestion to take the id of a patch from the xml
output to use it with --patch 'hash xxx'. This seems to be a terrible
UI. Even the first characters of a patch name are longer and possibly
not distinctive enough.
These patch numbers could also be used to annotate the lines of a file.
The first few columns show the number of the patch. This works well in
many cvs viewers.

The drawback of this numbering is, that the user has to understand the
temporary nature of these numbers. They can only be used for short and
( with care ) medium term manipulation of a single repo.

By coincidence i already tried something like this. I converted a svn
repo to darcs with a very simple script. The patch name is just the
output of svn log. Every name begins with 'rxxxx', the subversion
revision number.

Regards
  Nils Decker

-- 
Nils Decker <ndecker at gmx.de>