Re: [PATCH 3/5] shell32: use xdg well known directories for shell redirection (try 4)

[Alexandre Julliard]

"Lei Zhang"  writes:
> This patch attempts to set the shell folders using xdg well known
> directories via shell redirection. If any part of this fails, the rest
> of SHELL_RegisterShellFolders() should fill in the gaps.

This will still break if the user changes the drives config. I don't
understand why you don't want to simply use the symlinks; why should the
dirs be handled differently when they are obtained from the xdg config
instead of from the current hard-coded values?

[PATCH 3/5] shell32: use xdg well known directories for shell redirection (try 4)

[Lei Zhang]

On Tue, Mar 11, 2008 at 3:39 AM, Alexandre Julliard  wrote:
> "Lei Zhang"  writes:
> > This patch attempts to set the shell folders using xdg well known
> > directories via shell redirection. If any part of this fails, the rest
> > of SHELL_RegisterShellFolders() should fill in the gaps.
>
> This will still break if the user changes the drives config. I don't
>  understand why you don't want to simply use the symlinks; why should the
>  dirs be handled differently when they are obtained from the xdg config
>  instead of from the current hard-coded values?
>
> --
>  Alexandre Julliard
>  [e-mail]

We ran into a problem with Picasa and directory scanning. It looks
like an issue with FindFirstChange/FindNextChange and overlapping
directories, so I'm trying to use shell redirection to avoid the
problem. Perhaps we should fix the problem with
FindFirstChange/FindNextChange rather than avoid it.

I'll resubmit the patches without the shell redirection code.

[PATCH 3/5] shell32: use xdg well known directories for shell redirection (try 4)

[Alexandre Julliard]

"Lei Zhang"  writes:
> We ran into a problem with Picasa and directory scanning. It looks
> like an issue with FindFirstChange/FindNextChange and overlapping
> directories, so I'm trying to use shell redirection to avoid the
> problem. Perhaps we should fix the problem with
> FindFirstChange/FindNextChange rather than avoid it.

We should at least understand what the problem is. And if it turns out
that we have to use shell redirection, then we want to do that in all
cases, it shouldn't be specific to the xdg case.
> I'll resubmit the patches without the shell redirection code.

Thanks.