On 8/9/06, Nico Kadel-Garcia <nkadel@comcast.net> wrote:
> Garrett Rooney wrote:
> > On 8/7/06, Nico Kadel-Garcia <nkadel@comcast.net> wrote:
> >> Garrett Rooney wrote:
> >>> On 8/5/06, Nico Kadel-Garcia <nkadel@comcast.net> wrote:
> >>>> I wrote about this before: the "svnadmin hotcopy" tool does not
> >>>> successfully replicate symlinks in the repositories.
> >>>>
> >>>> Someone sent me a nice patch for the "repos.c" file, which added
> >>>> the capability of duplicating symlinks, but I also had to add a
> >>>> patch to io.c in order to actually have the function called as the
> >>>> walk through the directories lists files and directories to move.
> >>>>
> >>>> I'm attaching a copy of the patch against the current trunk for
> >>>> subversion, I'll also include it in the issue tracker page I
> >>>> created.
> >>>
> >>> The patch seems mostly reasonable, but the bit where you fprintf to
> >>> stderr in the case of something you don't understand is just wrong.
> >>> We can't just spew stuff to stderr in a library function like that,
> >>> there's no way to tell if anyone's even looking at stderr, this
> >>> could be embedded in a GUI app or something. If we NEED a way to
> >>> warn without erroring, we have to do it via some sort of
> >>> callback/baton pair, otherwise treat it as an error and return an
> >>> svn_error_t just like any other problem. Or just look at the other
> >>> potential cases and figure out if they need to be handled, if not
> >>> just ignore them.
> >>
> >> A reported error is *fine* with me! I wasn't sure what best to
> >> inject, and I'm always ware of functions relying on being called
> >> only with correct arguments. There should have been a reported error
> >> in the first place, rather than a silently failed function. Do you
> >> know the best syntax for that return?
> >
> > Just return an svn_error_t, like every other error return in that
> > file...
>
> My preliminary attempt to add such an error report is not working well: I'm
> afraid that I would have to travel back up the pyramid of functions to
> really establish some error reporting. Is anyone else more comfortable doing
> that?
I'd suggest bringing questions like that to the dev list, if you tell
us what's going wrong with your patch, perhaps we can help you fix it.
-garrett
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Aug 9 22:50:25 2006