Re: [evolution-patches] 31745, hacky workaround for shell uri bugs

[Not Zed <notzed ximian com>]

committed


On Mon, 2003-06-16 at 17:14, Not Zed wrote:
> Plenty of discussion in the bug in question.
> 
> The problem is, the shell gets something like this from camel:
> 
> physical = imap://user host/%23mh/folder
> path = /#imap/folder
> name = folder
> 
> and uses the 'path' part directly in its own uri formation, rather than
> the encoded uri path.
> 
> e.g. e-shell-view.c:switch_on_folder_tree_click()
> 
> uri = g_strconcat(E_SHELL_URI_PREFIX, path, NULL);
> 
> thus forming a non-encoded/invalid uri.  its actually the 'view
> fragment' (?) code (in display_uri() of the same file) that stops this
> working though (since the path is used as a strcmp in most places
> 'safely').
> 
> This is a quick and dirty workaround, since the problem seems to be
> pretty major/indemic in the shell code.  However, the main caveat would
> be that if we put the patch in, it will affect users when we later fix
> it properly (filters, and imap cache, etc).
> 
> It just strips any '#'s from the namespace component which is exported
> via uri's and paths (not heavily tested, but i could access mail using
> it).  It could create a namespace clash, but not very likely i'd say,
> and so could any other scheme.
> 
> I moved the bug to 1.4.x territory because it makes evolution useless
> for some users.
>