You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Karl Fogel <kf...@newton.ch.collab.net> on 2002/07/03 14:45:29 UTC

Re: svn commit: rev 2406 - trunk/subversion/libsvn_ra_dav

striker@tigris.org writes:
> * subversion/libsvn_ra_dav/commit.c
> 
>   (commit_add_dir, commit_add_file): escape the uri before passing it to
>     neon.  This fixes adding with history for dirs/files that contain
>     spaces.

Ummm.  Hmmm.  Sander, have you read over all of

   http://subversion.tigris.org/issues/show_bug.cgi?id=494

?

The quick summary is: we're not going to do URI escaping inside the
svn libraries.  Instead, the libraries are going to accept pre-escaped
URIs, and it's up to the client itself (the user client, I mean, not
libsvn_client) to make sure the URI is escaped without having lost any
meaning.

Whether the client does this by insisting that the user type "%20"
instead of " ", or by escaping spaces itself, is outside the scope of
issue #494, and is the topic of another thread, so I won't go into it
here.  But either way, libsvn_ra_dav should not be performing any
escaping internally.

Yes, this implies that we will also be storing URIs in escaped form,
for example in the entries file.

So, I think this change needs to come out.  The real fix happens at a
higher layer, and I'm working on issue #494 right now, so you should
see it soon.

I'm not reverting just yet, want to get your (and others') thoughts
first.

Thanks,
-Karl

> Modified: trunk/subversion/libsvn_ra_dav/commit.c
> ==============================================================================
> --- trunk/subversion/libsvn_ra_dav/commit.c	(original)
> +++ trunk/subversion/libsvn_ra_dav/commit.c	Wed Jul  3 09:02:24 2002
> @@ -746,6 +746,7 @@
>           header given to COPY is simply the wr_url that is already
>           part of the child object. */
>        copy_src = svn_path_join(bc_url.data, bc_relative.data, dir_pool);
> +      copy_src = svn_path_uri_encode(copy_src, dir_pool);
>  
>        /* Have neon do the COPY. */
>        status = ne_copy(parent->cc->ras->sess,
> @@ -933,6 +934,7 @@
>           header given to COPY is simply the wr_url that is already
>           part of the file_baton. */
>        copy_src = svn_path_join(bc_url.data, bc_relative.data, file_pool);
> +      copy_src = svn_path_uri_encode(copy_src, file_pool);
>  
>        /* Have neon do the COPY. */
>        status = ne_copy(parent->cc->ras->sess,
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: svn-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: svn-help@subversion.tigris.org

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

RE: svn commit: rev 2406 - trunk/subversion/libsvn_ra_dav

Posted by Sander Striker <st...@apache.org>.
> From: Karl Fogel [mailto:kfogel@newton.ch.collab.net]
> Sent: 03 July 2002 16:45

> striker@tigris.org writes:
> > * subversion/libsvn_ra_dav/commit.c
> > 
> >   (commit_add_dir, commit_add_file): escape the uri before passing it to
> >     neon.  This fixes adding with history for dirs/files that contain
> >     spaces.
> 
> Ummm.  Hmmm.  Sander, have you read over all of
> 
>    http://subversion.tigris.org/issues/show_bug.cgi?id=494
> 
> ?
> 
> The quick summary is: we're not going to do URI escaping inside the
> svn libraries.  Instead, the libraries are going to accept pre-escaped
> URIs, and it's up to the client itself (the user client, I mean, not
> libsvn_client) to make sure the URI is escaped without having lost any
> meaning.
> 
> Whether the client does this by insisting that the user type "%20"
> instead of " ", or by escaping spaces itself, is outside the scope of
> issue #494, and is the topic of another thread, so I won't go into it
> here.  But either way, libsvn_ra_dav should not be performing any
> escaping internally.
> 
> Yes, this implies that we will also be storing URIs in escaped form,
> for example in the entries file.

I didn't realize that it would extend to the entries file.
 
> So, I think this change needs to come out.  The real fix happens at a
> higher layer, and I'm working on issue #494 right now, so you should
> see it soon.
> 
> I'm not reverting just yet, want to get your (and others') thoughts
> first.

Just for the record.  I agreed with Karl on IRC that we should revert
(preferably when the issue is solved at a higher level, ie. #494 completion).
The reason for not replying here was simply because mail was down
for @apache.org.

Sander


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