On Mon, Aug 01, 2005 at 08:17:52AM -0400, David Roundy wrote:
> On Mon, Aug 01, 2005 at 01:36:12PM +0200, Lele Gaifax wrote:
> > >>>>> "David" == David Roundy <[EMAIL PROTECTED]> writes:
> > 
> >     >> A patch cannot be deleted, unless it also appears in any given
> >     >> remote repository, or some options, such as "--force", are
> >     >> given.
> > 
> >     David> I like this idea! ... I'm imagining this change in
> >     David> conjunction with the introduction of a more dangerous
> >     David> "obliterate-like" command, which would be closer to todays
> >     David> unpull (in its dangerousness).
> > 
> > Isn't a single command enough? With "--force" it wouldn't perform the
> > check on known related repositories before unconditionally dropping
> > the patch...
> 
> That would be enough, but I have two objections.  First, I don't like the
> idea of combining a "safe" command and a "dangerous" command into a single
> command.  This is the reason I haven't been willing to combine unrecord
> with unpull either--one is safe and the other unsafe.  Secondly, the safe
> behavior make sense for "unpull" but not for "drop" or "obliterate", and
> "unpull --force" is even more confusing a command for deleting a patch than
> "unpull" is.

I really don't know why is there such a problem with user deleting any
of his data, if he really wants.  I mean, all of this just complicates
both the interface and implementation in exchange for a tiny bit of
security.  You really can't protect users from themselves, no matter what
you do.  Suppose he does 'rm -rf my_lovely_repo', how would you go about
that?

As far as I can tell, there have always been much more complaints about
the naming of unpull than dangerousness of that command.  So I suggest
you just rename the command to 'drop' and print a big warning, like this:

This command is DANGEROUS, it deletes your patches permanently.
Are you REALLY SURE you want to do this (yes/no)?

And perhaps then force the user to type _exactly_ 'yes' to proceed.

This is, in my humble opinion, much better than to add new commands
or change semantics of already existing commands in unexpected ways.

-- 
Ivan Stankovic, [EMAIL PROTECTED]

_______________________________________________
darcs-users mailing list
[email protected]
http://www.abridgegame.org/mailman/listinfo/darcs-users

Reply via email to