You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Greg Stein <gs...@lyra.org> on 2003/05/14 02:56:24 UTC

Re: svn commit: rev 5929 - trunk/subversion/libsvn_wc

On Tue, May 13, 2003 at 08:28:07PM -0500, sussman@tigris.org wrote:
>...
> @@ -776,6 +763,14 @@
>            (SVN_ERR_WC_OBSTRUCTED_UPDATE, NULL,
>             "failed to add dir '%s': \nobject of the same name is already "
>             "scheduled for addition", path);
> +
> +      /* Immediately create an entry for the new directory in the parent.
> +         Note that the parent must already be either added or opened, and
> +         thus it's in an 'incomplete' state just like the new dir.  */      
> +      tmp_entry.kind = svn_node_dir;
> +      SVN_ERR (svn_wc__entry_modify (adm_access, db->name, &tmp_entry,
> +                                     SVN_WC__ENTRY_MODIFY_KIND,
> +                                     TRUE /* immediate write */, pool));

Don't you want to set the 'incomplete' flag here?

Oh. This is in the parent dir, huh? The incomplete flag is only mentioned in
THIS_DIR ?

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: svn commit: rev 5929 - trunk/subversion/libsvn_wc

Posted by Ben Collins-Sussman <su...@collab.net>.
Greg Stein <gs...@lyra.org> writes:

> On Tue, May 13, 2003 at 08:28:07PM -0500, sussman@tigris.org wrote:
> >...
> > @@ -776,6 +763,14 @@
> >            (SVN_ERR_WC_OBSTRUCTED_UPDATE, NULL,
> >             "failed to add dir '%s': \nobject of the same name is already "
> >             "scheduled for addition", path);
> > +
> > +      /* Immediately create an entry for the new directory in the parent.
> > +         Note that the parent must already be either added or opened, and
> > +         thus it's in an 'incomplete' state just like the new dir.  */      
> > +      tmp_entry.kind = svn_node_dir;
> > +      SVN_ERR (svn_wc__entry_modify (adm_access, db->name, &tmp_entry,
> > +                                     SVN_WC__ENTRY_MODIFY_KIND,
> > +                                     TRUE /* immediate write */, pool));
> 
> Don't you want to set the 'incomplete' flag here?
> 
> Oh. This is in the parent dir, huh? The incomplete flag is only mentioned in
> THIS_DIR ?

Yup.  Our universal rule of entries files is:  a parent directory's
pointer to a subdir is nothing but a name, a file-kind, and possibly a
schedule flag.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Repeatability Files

Posted by Jonathan Smith <jo...@members.limitless.org>.
In our project, we test the output's repeatability from each change to the 
next.  The ouput consists of maybe 10 meg.  In order to facilitate 
conflict resolution, I need to have these files available for the most 
recent version of the code to compare my version with the conflicts 
resolved against.  As this output *can* be generated from the compiled 
version of the code in the repository, this output, strictly speaking, 
shouldn't be revision controlled.

For example:
repository has version n
I checkout version n
Harry commits version n+1
My change is repeatable
I update to version n+1 with my changes
I need the output of version n+1 to compare to, I only have version n
Recompiling version n+1 without my changes is virtually impossible without 
	a second checkout of the sourcecode.

My thought is, fine.  Just commit the output files...  store them in a 
separate top level directory that can be disregarded; however, link them 
in where they belong in the repository.  I don't remember what this is 
called, possibly externals, but I remember reading about it in the 
documentation.  But, thats about 20 lines of change simply by re-running 
the program (time-date stamps) and hundreds of thousands if we actually 
change the output.  Can we reduce this with a meta data tag to specify to 
only hold N revisions in history?  I'd be agreeable to N=1, though i'd 
prefer N=5 or N=10.

I'm sure that this is far, far, far easier than my last request.

I'm *really* starting to like the meta data.  Power to the users!

j.


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org