Re: Mailcap, tmpfiles and master processes such as browsers
On Sat, Feb 02, 2002 at 11:18:32PM +0100, Jeroen Dekkers wrote:
> On Sat, Feb 02, 2002 at 09:42:20PM +0100, Erich Schubert wrote:
> > > Why does the 2nd galeon instance remove a tmpfile that it did not
> > > create? That looks like the problem.
> > The problem is that the caller cannot know when it can remove the
> > tmpfile. Ususally you remove the temp file, when the process you started
> > terminates. This fails and the tempfile gets removed too soon, when the
> > launched process just "triggers" another process and then terminates.
> Galeon should implement something like emacsclient does. The 2nd
> instance should terminate when the window closes or when the user
> tells it to do so. This is the way emacsclient works (the emacsclient
> process terminates when you either tell it to do so with C-x # or when
> you kill the buffer).
Not quite feasible for web browsers, I think. emacsclient is called
instead of a real editor and is expected to have the same semantics
(exit after editing is done and edited file can be used).
Web browsing may not be "done" at any time, we might let the user follow
links in the loaded page and still let them continue in their mail
reader. Technically we could require that the window/tab which loaded
the initial page would have to be closed. It would be inconsistent with
starting a fresh browser, anyway.
The real fix (as mentioned in this thread already) is of course to make
galeon wait until the main process has opened the file before exiting.
It could even return a sensible status then (file found/not found).
Andreas Bombe <email@example.com> DSA key 0x04880A44