[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

diff_repos_wc question

From: Neels J. Hofmeyr <neels_at_elego.de>
Date: Fri, 03 Oct 2008 06:12:51 +0200

I am confused... when I look in the code, I find

subversion/libsvn_client/diff.c:1226
[[[
/* Perform a diff between a repository path and a working-copy path.

   PATH1 may be either a URL or a working copy path. PATH2 is a
   working copy path. REVISION1 and REVISION2 are their respective
   revisions. If REVERSE is TRUE, the diff will be done in reverse.
   If PEG_REVISION is specified, then PATH1 is the path in the peg
   revision, and the actual repository path to be compared is
   determined by following copy history.

   All other options are the same as those passed to svn_client_diff4(). */
static svn_error_t *
diff_repos_wc(const char *path1,
              const svn_opt_revision_t *revision1,
              const svn_opt_revision_t *peg_revision,
              const char *path2,
              const svn_opt_revision_t *revision2,
              svn_boolean_t reverse,
              svn_depth_t depth,
              svn_boolean_t ignore_ancestry,
              const apr_array_header_t *changelists,
              const svn_wc_diff_callbacks3_t *callbacks,
              struct diff_cmd_baton *callback_baton,
              svn_client_ctx_t *ctx,
              apr_pool_t *pool)
]]]

and

:1363
[[[
/* This is basically just the guts of svn_client_diff[_peg]3(). */
static svn_error_t *
do_diff(const struct diff_parameters *diff_param,
        const svn_wc_diff_callbacks3_t *callbacks,
        struct diff_cmd_baton *callback_baton,
        svn_client_ctx_t *ctx,
        apr_pool_t *pool)
{
  struct diff_paths diff_paths;

  /* Check if paths/revisions are urls/local. */
  SVN_ERR(check_paths(diff_param, &diff_paths));

  if (diff_paths.is_repos1)
    {
      if (diff_paths.is_repos2)
        {
          return diff_repos_repos(diff_param, callbacks, callback_baton,
                                  ctx, pool);
        }
      else /* path2 is a working copy path */
        {
          return diff_repos_wc(diff_param->path1, diff_param->revision1,
                               diff_param->peg_revision,
                               diff_param->path2, diff_param->revision2,
                               FALSE, diff_param->depth,
                               diff_param->ignore_ancestry,
                               diff_param->changelists,
                               callbacks, callback_baton, ctx, pool);
]]]

So there is code that decides to do a diff between a repos path and a
working copy. But I can't seem to diff a repository revision against a
working copy with local modifications from the commandline.

I am trying to diff path A/B/E at rev 2 against the locally modified working
copy of A/B3/E (A/B3 being a modified copy of A/B):

[[[
$ pwd
[...]/A/B3
$ svn diff file:///[...]/A/B/E_at_2 E
subversion/svn/diff-cmd.c:282: (apr_err=200007)
svn: Target lists to diff may not contain both working copy paths and URLs

$ cd E
$ pwd
[...]/A/B3/E
$ svn diff file:///[...]/A/B/E_at_2
subversion/libsvn_client/diff.c:900: (apr_err=195002)
svn: Not all required revisions are specified

$ pwd
[...]/A/B3/E
$ svn diff -r 2 ../../B/E
subversion/libsvn_client/diff.c:836: (apr_err=150000)
svn: '../../B/E' is not under version control
# oh, ok, it is removed on HEAD. Restore...
$ svn up -r 2 ../../B
A ../../B/E
A ../../B/E/alpha
A ../../B/E/newfile
A ../../B/E/beta
Updated to revision 2.
$ svn diff --revision 2 ../../B/E
$ # No output at all, but there are differences
$ ls
alpha beta bumm newfile
$ ls ../../B/E/
alpha beta newfile
$ cat newfile
This is the file 'newfile'.
Last changed in '$Revision: 6 $'.
karl
$ cat ../../B/E/newfile
This is the file 'newfile'.
Last changed in '$Revision: 2 $'.
$ svn ci -m m
$ svn diff --revision 2 ../../B/E
$
]]]

The subversion book says on the diff command: "OLD-TGT and NEW-TGT may be
working copy paths or URL[@REV]." I assumed it also means that OLD-TGT may
be a repository URL while NEW-TGT may be a working copy path.

It furthermore says:
[[[
--revision N
    The client compares TARGET_at_N against the working copy.
]]]

in this context:

[[[
If TARGET is a working copy path, the default behavior (when no --revision
option is provided) is to display the differences between the base and
working copies of TARGET. If a --revision option is specified in this
scenario, though, it means:

--revision N:M
    The server compares TARGET_at_N and TARGET_at_M.

--revision N
    The client compares TARGET_at_N against the working copy.

If the alternate syntax is used, the server compares URL1 and URL2 at
revisions N and M, respectively. If either N or M is omitted, a value of
HEAD is assumed.

By default, svn diff ignores the ancestry of files and merely compares the
contents of the two files being compared. If you use --notice-ancestry, the
ancestry of the paths in question will be taken into consideration when
comparing revisions (i.e., if you run svn diff on two files with identical
contents but different ancestry, you will see the entire contents of the
file as having been removed and added again).
]]]

Which is my third diff command shown above. Well, it probably means to say
that it compares TARGET_at_N against the working copy at TARGET. To me it reads
though that it compares TARGET_at_N to the working copy at `pwd`.

The general sound of the diff help is that any diff combination is possible.
There is code that looks like that's the case. The subversion book is at
least ambiguous on the question.

Help anyone?

Thanks,
~Neels

-- 
Neels Hofmeyr -- elego Software Solutions GmbH
Gustav-Meyer-Allee 25 / Gebäude 12, 13355 Berlin, Germany
phone: +49 30 23458696  mobile: +49 177 2345869  fax: +49 30 23458695
http://www.elegosoft.com | Geschäftsführer: Olaf Wagner | Sitz: Berlin
Handelsreg: Amtsgericht Charlottenburg HRB 77719 | USt-IdNr: DE163214194

Received on 2008-10-03 06:13:19 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.