On Monday 03 November 2003 4:03 pm, Kayvan A. Sylvan wrote:
> On Mon, Nov 03, 2003 at 01:21:07PM +0000, Angus Leeming wrote:
> > Kayvan has implemented the most attrocious hack that I suspect is the
> > real cause of our problems under mingw. This is from os_win32.C:
> >
> > string os::external_path(string const & p) {
> > string dos_path=p;
> > if (is_absolute_path(p)) {
> > char dp[255];
> > cygwin_conv_to_full_win32_path(p.c_str(), dp);
> > dos_path=subst(dp,'\\','/');
> > }
> > return dos_path;
> > }
> >
> > Cygwin is clever enough to untangle the mess. I'd prefer that we fixed
> > things properly but am still unclear why Kayvan found it to be
> > necessary in the first place. (He did explain, but I failed to follow
> > the argument.)
>
> The problem is with the latest Cygwin tetex. The old one handled
> both the Win32 style and the Unix style paths correctly.
>
> The latext Cygwin tetex only handles the Unix-style paths. Therefore,
> the conversion to Win32 paths is unnecessary and counterproductive.
>
> I have no problems with graphics (rendering on screen or running
> through LaTeX) with my atrocious hack.
Sorry about my language. It was meant to sound pejorative.
Unfortunately, what works for you appears to break things for those running
the Win32 port. Hence my suggestion: wrap the call to latex inside a script
if such manipulation is needed.
I'm sure that we can do this painlessly using some configure magic.
Angus
