Re: [PATCH] Improve gnome_vfs_xfer_delete_items_common() error handling

[Alexander Larsson <alexl redhat com>]

On Tue, 2006-12-26 at 21:46 +0100, Christian Neumair wrote:
> The attached patch makes the deletion code handle errors during file
> info fetching.
> 
> The progress phase is quite fragile, even after this patch has been
> applied :(.
> 
> The deletion helper is used for deletion operations, and for cleanup
> during a merged copy (i.e. a move operation that conflicted, or a move
> between different FS). It sounds like a bad idea to hop forth and back
> between the progress phases (i.e. CLEANUP -> DELETESOURCE (error
> handling) -> CLEANUP (normal progress) -> ...), but essentially we'll
> see an unannounced CLEANUP->DELETESOURCE hop (in the call_often in
> remove_directory/file) when called in CLEANUP mode, since we don't never
> do a call_progress.
> 
> I am also aware that for multiple URIs in the CLEANUP case, we'll never
> get a CLEANUP status again, so thephase seen by the error handler
> depends on whether we're eating the first URI or subsequent ones.
> 
> I have no clue how we should handle this (the current phase bit just
> doesn't seem to be appropriate), but the most important thing seems to
> be to make users aware of any file operation failure, especially in the
> VFAT case, not the details of the progress info (aside from the URIs).

Yeah, the phase stuff is really undocumented, strange and fragile.
Hopefully we'll be able to totally replace xfer in gvfs.

Anyway, i think that looks good to commit. Reporting errors can't be
wrong.

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 Alexander Larsson                                            Red Hat, Inc 
                   alexl redhat com    alla lysator liu se 
He's a jaded umbrella-wielding shaman from a doomed world. She's a 
cold-hearted belly-dancing barmaid on the trail of a serial killer. They fight 
crime!