On 10/5/07, Bert Huijben <bert@vmoo.com> wrote:
> [Another version of this mail will follow in a couple of hours; it is
> waiting for moderation as I used another mail address]
> > -----Original Message-----
> > From: Karl Fogel [mailto:kfogel@red-bean.com]
> > Sent: donderdag 4 oktober 2007 8:08
> > To: dev@subversion.tigris.org
> > Subject: We should bump WC format version number for 1.5.
> >
> > When I started writing this mail, it was a question: "Should we bump
> > the wc format number for 1.5?"
> >
> > Now I've done some investigation, and I think it's a statement
> > instead. Despite my ardent desire that we not bump the format number,
> > we may have to. (There *might* be a way to conditionalize it; search
> > for "escape hatch" later in this mail to read about that.)
>
> When I look through this thread I think we forget two other important
> feature of the 1.5 release: We started using merge tracking & changelists.
>
>
> Merge tracking will give 1.4 clients+users a hard time on updates and merges
> as they don't have any builtin merge support for the svn:mergeinfo property.
>
> (See my other mail: 1.5 clients update this property on merges, copying,
> etc. and merge this property automatically.. 1.4 clients can only use it as
> an ordinary property)
>
>
> Looking through svn_entries I found these other wc additions since 1.4.X:
> - changelist
> File registration on a changelist is reset
Yes, I guess that would be a problem.
> - keep_local (impacts directories only)
> svn rm --keep-local dir
> Won't keep directory
That too would be a problem.
> - working-size
> svn status won't see changes when the timestamp has not changed but the file
> size has.
This is not a problem: the 1.5 client just won't perform any better at
detecting changes than the 1.4 client would... It's not a regression
or a real problem, but definitely a pity.
> ===
> All these items break if 1.4 and 1.5 clients are used on the same working
> copy.
Yes. Thanks for completing the list.
The list above makes me think we should require wc-format upgrade on
all write operations, but only if the user explicitly chooses to do
so. (Better said: we should require the right format for all write
ops, erroring if it's not. Then 'svn cleanup --some-flag' could be
used to do an explicit upgrade...)
bye,
Erik.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Oct 5 15:14:07 2007