> From: sussman@collab.net [mailto:sussman@collab.net]
> Sent: 22 October 2002 22:40
[...]
>> I'll tell you why I ask so suddenly. I just started working on issue
>> #933, and pretty soon found all roads leading to `import' simply being
>> a `cp' that works on non-versioned items.
>>
>> Before I go any farther, we need to figure out what we want. I'm
>> beginning to think what we want is for 'svn cp' to Do The Right Thing
>> when copying non-versioned data to a URL dest.
>
> I agree that this is a great approach, at least in terms of internal
> implementation.
>
> But I think that the concept of 'import' should *not* magically
> absorbed into the idea of 'cp'. Here's why:
>
> 1. We already have 4 'svn cp' use-cases. Let's not add a 5th.
> Talk about brain clutter.
>
> 2. The concept of 'import' and 'copy' don't necessarily blend for
> everyone. When I do an import, for example, I don't think of it
> as a copy; I think of it as a shortcut to 'svn checkout; svn add
> -R; svn commit'.
>
> 3. An 'import' is usually a one-time special event. The concept is
> friendly to CVS users, which is important. I don't want the
> concept to go away just because it *can* go away. It's the same
> reason we didn't blend 'checkout, update, switch' all into a
> single 'fetch' command. It's a matter of walking the fine-line
> between too many subcommands (like bitkeeper, so others tell
> me), and too *few* subcommands.
>
> Anyway, I guess I'm just suggesting that you keep the same old 'svn
> import' command, and give it the copy-like syntax:
>
> svn import from-dir to-URL
>
> ... and under the hood, if you want to call a variant of
> wc_to_repos_copy, then great!
+1
Sander
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Oct 22 23:56:54 2002