On Wed, 2009-08-12 at 16:07 +0300, Daniel Shahaf wrote:
> Julian Foad wrote on Wed, 12 Aug 2009 at 13:47 +0100:
> > Hi Daniel. This all looks great but I think it makes an unintended
> > behaviour change. If 'foo' is non-existent, then:
> >
> > Running svn v1.6:
> , not in a WC:
> > [[[
> > $ svn up foo
> > Skipped 'foo'
> > ]]]
> >
> > Running your version, not in a WC:
> > [[[
> > $ svn up foo
> > Skipped 'foo'
> > Summary of conflicts:
> > Skipped paths: 1
> > ]]]
> >
> > That's good.
> >
>
> ...
>
> > Running your version, in a WC:
> > [[[
> > $ svn up foo
> > At revision 38693.
> > ]]]
> >
> > That's the unintended change. The failure to print "Skipped 'foo'" is
>
> It's not a change --- svn16 has the same output when inside a WC.
Not on my system, using svn built from the 1.6.x branch on 2009-07-15.
What system and version are you testing on?
> > serious. The fact that it prints "At revision ..." is not so serious: we
> > could accept that (in addition to a "Skipped" message) because it is
> > analogous to the case of updating a versioned child of a versioned
> > directory.
> >
>
> I think in this case the file *really* isn't skipped --- for example, if
> the wc is at r4 and 'foo' was created at r5, then 'svn up foo' works.
> (I'm not saying this is the way it *should* work. But this is how it
> *does* work in 1.6.)
If 'foo' was created at r5, then 'svn up foo' would print
[[[
A foo
Updated to revision 5.
]]]
not just
[[[
At revision 5.
]]]
That's OK, and that's not what I'm talking about. I'm talking about the
case where 'foo' doesn't exist either locally or in the repository at
any revision.
- Julian
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2382897
Received on 2009-08-12 15:42:26 CEST