subversion-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Julian Foad <>
Subject Re: reverse-merge fails to undo replacements
Date Thu, 05 Jun 2014 21:25:45 GMT
Stefan Sperling wrote:
> Paths which were replaced are not reverse-merged.
> Is this intended behaviour or a bug?

The first key fact here is those replacements involve no change of content:

  $ svn log -r42758
  r42758 | [...]
  Changed paths:
     R /trunk/docs/core/howto/listings/logger (from /branches/log-booyah-6750-4/docs/core/howto/listings/logger:42757)

  $ svn diff -c 42758 ^/trunk/docs
  [no output]

The second key fact is that the replacement is related to the replaced node. That is, I suppose,
the reason that even the "notice ancestry" mode of diff shows no replacement:

  $ svn diff -c 42758 --notice-ancestry ^/trunk/docs
  [no output]

I'm not sure exactly what the relationship is, in terms of which was copied from what at what
revision -- I haven't checked -- but there's some copied-from relationship there.

The changes that "merge" applies are generally meant to be the same 
changes that "diff --notice-ancestry" shows. In that respect, "merge" is
working consistently with "diff".

As the merge is behaving like diff, I can't see a bug here unless there is some further inconsistency
to be found.

I haven't tried the corresponding forward merge, but I presume it would also do nothing, just
like the reverse merge.

I admit it's odd that "log -v" shows a replacement but "diff --notice-ancestry" doesn't; but
that's a rather different issue.

- Julian

> svn co svn:// wc
> cd wc
> $ svn merge -c -42758 .
> --- Reverse-merging r42758 into '.':
> D    twisted/topfiles/6750.feature
> $ svn log -r42758
> ------------------------------------------------------------------------
> r42758 | wsanchez | 2014-06-03 07:25:58 +0200 (Tue, 03 Jun 2014) | 8 lines
> Changed paths:
>    R /trunk/docs/core/howto/listings/logger (from 
> /branches/log-booyah-6750-4/docs/core/howto/listings/logger:42757)

View raw message