Karl Fogel wrote:
> Should commit accept all possible values of --depth?
>
> Right now, commit takes -N. The purpose of that is to allow someone
> to commit a property change on a directory without committing changes
> to files or subtrees under that directory.
>
> (Most other commands treat -N like --depth=files, but commit is
> special: 'svn commit -N DIR' doesn't commit modified files in DIR, it
> just commits any propchanges on DIR. This is not what you'd expect
> from reading svn_client_commit4()'s doc string, which currently says:
> "If @a recurse is false, subdirectories of directories in @a targets
> will be ignored". My testing indicates that that doc string is just
> mistaken, and I'll fix it appropriately depending on the outcome of
> this thread.)
>
> Let's consider what full --depth behavior might look like for commits:
>
> --depth=empty: Like today's -N.
>
> --depth=files: Like depth=empty, plus also commits any modified
> files directly inside a named directory target.
>
> --depth=immediates: Like depth=files, plus commits propchanges on
> any subdirectories of a named directory target.
>
> --depth=infinity: Full recursion, the default.
>
> Now the $64,000 question: are depth=files and depth=immediates really
> useful? I can see use cases for depth=empty and depth=infinity, but
> what would we really gain by supporting the other two?
>
What is the reason not to include it? Technical objections? Scope
limiting for 1.5?
I'm sure that even if we can't think of a specific usage scenario, some
users will try it and then be confused why --depth for commit is not
behaving the same way as for update or status.
Lieven
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Jul 1 18:29:34 2007