Re: Last lonely little test failure for the Windows Ruby bindings

Hi Joe,

In <ae6cb1100703242138o6e34716cma4899e47d8408d09@mail.gmail.com>
  "Re: Last lonely little test failure for the Windows Ruby bindings" on Sat, 24 Mar 2007 21:38:39 -0700,
  "Joe Swatosh" <joe.swatosh@gmail.com> wrote:

> > > I am curious about the branching based on windows? in
> > > assert_translated_file_eol however. Why do the svn_wc_translate_file
> > > functions return LF instead of CRLF?
> >
> > I don't know why and whether is it a bug or not. We need to
> > find out it but I've skipped it. Sorry. Could you find out
> > it?
> >
>
> I started looking at this a little bit. It is misleading to me that
> Svn::Wc::translated_file doesn't forward to
> Svn::Ext::Wc::translated_file in the way that translated_file2 does.
> What are you trying to accomplish here?

Svn::Wc.translated_file just exists for backward
compatibility. We can use Svn::Ext::Wc.translated_file
instead. But there is a pool related
problem.

svn_wc_translated_file() makes a temporary file and returns
a file name of it. And the temporary file is deleted if it's
pool is destroyed. But the Ruby bindings can't control that
because returned file name doesn't have a reference of
temporary file's pool. So, I changed that the Ruby bindings
extract a content of temporary file, put it Tempfile and
destroy Subversion's temporary file. Now, temporary content
is managed under the Ruby bindings.

Thanks,

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