Re: Unable to read history on latest Cygwin

["Peter A. Castro" <doctor@xxxxxxxxxxxx>]

On Mon, 14 Dec 2009, Peter Stephenson wrote:

> "Peter A. Castro" wrote:
> > On Mon, 14 Dec 2009, Peter Stephenson wrote:
> > > On Mon, 14 Dec 2009 15:19:29 +0100
> > > "Thorsten Kampe" <thorsten@xxxxxxxxxxxxxxxx> wrote:
> > > > "Hardlinks on FAT [...] have been faked in
> > > > 1.5 by copying the file. The decision was made a couple of weeks back to
> > > > drop this fake since the application should know that the underlying
> > > > filesystem is not capable of creating hardlinks."
> > >
> > > I presume that means symlink() works OK?  We should probably be able to do
> > > something relatively straightforward like the following---using exactly the
> > > same code for symbolic links as hard links, most of the change is indenting
> > > code---though I've only checked it to the extent that it doesn't screw up
> > > somewhere that does support hard links.
> >
> > There is another option to consider here: not using link().
> >
> > The hist code has an alternative code path if link() is not available
> > (#ifdef HAVE_LINK), and it appears this is the only place in zsh code
> > which uses link().  I'm hacking configure to disable this usage on Cygwin
> > to see what, exactly, the difference in behaviour will be.
>
> I think that's rather more brittle: the code I've suggested (assuming it
> does the trick) doesn't require anything more than that link() and
> symlink() exist in the system library and covers any similar problem on
> other systems.  Hacking something specific for any system isn't usually
> desirable, plus all we've been told is that link() doesn't work on
> FAT32, not that it doesn't work under Cygwin.

I'll apply your change and see what it does, though I'm a little
skeptical considering that symlinks are an artificial construct under
Cygwin.  Really they are application-level Windows Short-cut files and an
application needs to "understand" that that is what it is.  Cygwin does
interpret them, for Cygwin apps, but there have been some issues with
using them.  A quick test will tell all.

--
Peter A. Castro <doctor@xxxxxxxxxxxx> or <Peter.Castro@xxxxxxxxxx>
	"Cats are just autistic Dogs" -- Dr. Tony Attwood