Frederic Melot wrote on Fri, 31 Oct 2008 at 17:57 +0100:
> Hello,
>
> you are totally right. My first concern was only on the "Copied:" headers
> for directories and I didn't change anything for "Added:" directories. I'll
> do it.
>
Okay. Note that there are a few other variants (e.g. whether the add is
of a file or of a dir, whether it is with/without history) -- watch your
step, then, don't break one while fixing the other ;)
> For the second point, I totally agree. But the code of the
> 'print_diff_tree' function is mainly triggerring 3 cases:
> 1) the copied files or folders, which update the string buffer and doesn't
> print it.
> 2) the files modifications, which update the string buffer and display it.
> It is the only code part in which the string buffer is displayed. This is
> why diffs on folders never appear.
> 3) the meta data modifications, a part which does not make use of the
What is "meta data"? Did you mean 'properties'?
> string buffer and prints directly the results
> As the string buffer was already not fully used I didn't use it. If we want
> to use it more, this function needs to be more deeper modified. Is it what
> you suggested?
>
I haven't actually suggested anything yet :)
The problem isn't that the stringbuf is underused; it is that its
accounting needs some fixes. Currently, sometimes the stringbuf is filled
but isn't printed; with your (previous) patch, it was not obvious that
there weren't cases where (parts of) the stringbuf would be printed twice.
Whether solving the accounting problem should be done by increasing the
use of stringbufs I leave for you to decide...
Thanks,
Daniel
> Best regards,
> Frederic
>
> Daniel Shahaf wrote:
> > Frederic Melot wrote on Tue, 21 Oct 2008 at 16:48 +0200:
> >
> > > Hello,
> > >
> > > here is a patch submission for issue #3292.
> > >
> > > Best regards,
> > > Frédéric Melot
> > >
> > >
> >
> > Frédéric, thanks for the patch. I have now tried it, and it still
> > doesn't print the "Added:" headers for added-without-history
> > directories. Could you look into that?
> >
> > Also, I'm worried that printing the header right away (as you did) could
> > lead to other problems: for example, how do you ensure that the header
> > isn't printed twice (in any case)? That code already isn't the easiest
> > to follow, and adding prints randomly scattered through won't make it
> > better.
> >
> > Thanks,
> >
> > Daniel
> > ------------------------------------------------------------------------
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
> > For additional commands, e-mail: dev-help_at_subversion.tigris.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-10-31 19:09:34 CET