[GRASS-dev] GRASS & QGIS: the future

[Glynn Clements]

Radim Blazek wrote:

> There is a problem in G_fatal_error():
> 
>     static int busy
>     if (busy)
>         exit(EXIT_FAILURE);
>     busy = 1;
> 
> second G_fatal_error() call always exits.
> 
> The busy check is there for the case when print error routine itself
> ends up with call to G_fatal_error()?

It's there to prevent infinite recursion, whether from the printing
code, a fatal error handler, or whatever.

> May be the busy variable made global and added a function to clear it?

That might be going a bit too far. OTOH, it can't be cleared prior to
calling the handler because it's intended to protect against the case
of a handler generating a fatal error.

One possibility would be to add explicit support for longjmp()ing out,
e.g. adding G_error_longjmp(jmp_buf env, int val) which would clear
the busy flag then perform the longjmp. That would avoid exposing the
internals more than is necessary.

Also, currently the global error handler (G_set_error_routine) is
called before the non-exclusive handlers (G_add_error_handler), so
those will never be called if the global handler lonjmp()s out. The
global error handler isn't limited to fatal errors, but is also used
for messages and warnings.

Perhaps we need a dedicated fatal error handler for this purpose? It
would be called immediately prior to the final exit() call. It would
not be allowed to call GRASS functions or to return. The busy flag
could be cleared before calling it.

-- 
Glynn Clements <glynn at gclements.plus.com>