Hi All,
Based on conversation in IRC, the plan is to do a filesystem move
to handle the automatic conflict resolution instead of working with
the API to try and recreate pristine contents of the temporary file
which would then be used to overwrite the conflicted file's contents.
(Temporary in this context means the files created when a conflict
happens. The .mine, .rOLDREV, or .rNEWREV.) If anyone has any
reasons why a move of one of those temporary files over the conflicted
file is a bad idea, please speak up. Consensus in IRC was that this
was the suggested approach.
Take care,
Jeremy
On 5/29/07, Jeremy Whitlock <jcscoobyrs@gmail.com> wrote:
> Hi All,
> Before I get to the question, let me give you some status on issue 2784:
>
> All unit tests are finished and failing as expected related to this issue.
> All revving of public APIs has been done.
> All necessary structures have been updated, when necessary, to
> allocate for the new argument.
> The resolved subcommand is now accepting and passing the passed accept argument.
>
> All that is left is the actual work which is where I'm proposing this:
> (Copied from IRC dialog)
>
> 17:10 < jeremyw> It appears that to get the contents of the file I
> want to use to resolve the conflict, I need to use the following:
> 17:11 < jeremyw> svn_wc_create_tmp_file2
> 17:11 < jeremyw> svn_io_copy_file
> 17:11 < jeremyw> Create the temporary file in tmp, copy it's contents
> to the file to be resolved.
> 17:12 < jeremyw> That is the way that merge.c does it.
> 17:12 < jeremyw> Anyone object, or have more to add?
> 17:12 < jeremyw> This is *only* an extension to the resolved
> subcommand so I'm not sure if loggy needs to be anymore involved than
> it already is.
> 17:16 < jeremyw> If you are unfamiliar but still want to be involved
> with answering this, libsvn_wc/merge.c line 447 should give you an
> idea of why I'm suggesting this approach.
>
> The proposed code above will go into libsvn_wc/adm_ops.c in function
> resolve_conflict_on_entry(). Can someone confirm my approach or pass
> insight as to the proper way to do this if my approach is erroneous?
> Based on my review of the code, I feel confident about the suggestion
> but with no response in IRC, I figured I would come here before just
> doing it my way. Thanks for the help.
>
> Take care,
>
> Jeremy
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu May 31 01:19:02 2007