[GRASS-dev] Can Rast_open_* call G_fatal_error()?

[Glynn Clements]

I wrote:

>  The fix would be simpler if G_open_cell_old() called G_fatal_error() on
>  errors. This would make life simpler for modules, which currently need to
>  check the return value. Off the top of my head, I don't know of a single
>  case where the caller actually tries to recover from G_open_cell_old()
>  failing. Doing so would also be a benefit to a small number of modules
>  which forget to check for errors.

Can anyone think of a reason *not* to implement this change in 7.0?

As it stands, just about every raster module does something like:

    infd = Rast_open_old(name, "");
    if (infd < 0)
	G_fatal_error(_("Unable to open input map <%s>"), name);

It would be simpler for just about everything if Rast_open_old() just
called G_fatal_error() itself on errors.

I can't think of a single case where the caller does something other
than call G_fatal_error() if opening an input map fails (or output
map, for that); although I did spot a couple of cases (e.g. 
v.vol.rst/user1.c) where the caller neglects to check the return
value.

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