You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Paul Burba <pt...@gmail.com> on 2008/01/25 16:13:17 UTC
Mergeinfo inheritance and mixed-rev WCs revisited
The Background:
The topic of mergeinfo inheritance within mixed-rev working copies was
first discussed here:
http://subversion.tigris.org/servlets/ReadMsg?listName=dev&msgNo=134295
In r28492 I made a change that prohibited inheritance across mixed-rev
boundaries.
Karl questioned the sanity of this change here,
http://subversion.tigris.org/servlets/ReadMsg?list=svn&msgNo=34070
and here,
http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=134338
-----
Karl,
Did you have a chance to think on this some more?
I'm somewhat ambivalent about taking working revisions into account
when doing mergeinfo inheritance. It is technically correct to do so
AFAICT, but in practical terms it is a headache whenever we have
mergeinfo anywhere other than branch roots. Worse I don't think
preventing inheritance across a mixed-rev path solves anything but
some far-out edge cases.
Isn't the larger issue about inheritable properties in general? They
are all fine and good in a WC with consistent working revisions. But
the moment mixed-revisions enter the picture we have to choose between
correctness and ease of use no?
Paul
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Mergeinfo inheritance and mixed-rev WCs revisited
Posted by Paul Burba <pt...@gmail.com>.
On Jan 25, 2008 3:23 PM, Karl Fogel <kf...@red-bean.com> wrote:
> "Paul Burba" <pt...@gmail.com> writes:
> > In r28492 I made a change that prohibited inheritance across mixed-rev
> > boundaries.
>
> r28942 (I knew which one you meant, no worry).
>
> > Karl questioned the sanity of this change here,
> >
> > Did you have a chance to think on this some more?
> >
> > I'm somewhat ambivalent about taking working revisions into account
> > when doing mergeinfo inheritance. It is technically correct to do so
> > AFAICT, but in practical terms it is a headache whenever we have
> > mergeinfo anywhere other than branch roots. Worse I don't think
> > preventing inheritance across a mixed-rev path solves anything but
> > some far-out edge cases.
> >
> > Isn't the larger issue about inheritable properties in general? They
> > are all fine and good in a WC with consistent working revisions. But
> > the moment mixed-revisions enter the picture we have to choose between
> > correctness and ease of use no?
>
> I agree with everything you say. But I haven't had much chance to
> think about it more, and I think it's okay to leave the code as it is
> for the moment -- this is an easy thing to tweak later, and if worse
> comes to worst and we have to ship with it too strict (and then relax
> later) that's better than the reverse. (Though I'd like to reconsider
> it before we ship.)
Sounds good, thanks for giving it a few brain cycles now. I'll raise
this again before the end.
I suspect a lot of folks see stuff like this as the minutia of merge
tracking, and I suppose it is! But IMHO inheritance may prove a bit
tricky to for users so I want to keep its behavior as understandable
and predictable as possible.
> In the meantime, you saw David Glasser's mail
> about this, right?:
>
> http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=134441
Yes, I saw that and responded in that thread.
Paul
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Mergeinfo inheritance and mixed-rev WCs revisited
Posted by Karl Fogel <kf...@red-bean.com>.
"Paul Burba" <pt...@gmail.com> writes:
> In r28492 I made a change that prohibited inheritance across mixed-rev
> boundaries.
r28942 (I knew which one you meant, no worry).
> Karl questioned the sanity of this change here,
>
> Did you have a chance to think on this some more?
>
> I'm somewhat ambivalent about taking working revisions into account
> when doing mergeinfo inheritance. It is technically correct to do so
> AFAICT, but in practical terms it is a headache whenever we have
> mergeinfo anywhere other than branch roots. Worse I don't think
> preventing inheritance across a mixed-rev path solves anything but
> some far-out edge cases.
>
> Isn't the larger issue about inheritable properties in general? They
> are all fine and good in a WC with consistent working revisions. But
> the moment mixed-revisions enter the picture we have to choose between
> correctness and ease of use no?
I agree with everything you say. But I haven't had much chance to
think about it more, and I think it's okay to leave the code as it is
for the moment -- this is an easy thing to tweak later, and if worse
comes to worst and we have to ship with it too strict (and then relax
later) that's better than the reverse. (Though I'd like to reconsider
it before we ship.) In the meantime, you saw David Glasser's mail
about this, right?:
http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=134441
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org