On Thu, 2010-01-21 at 00:17 +0100, Martin Kalbfuß wrote:
> The
> essential question is, why not to use g_assert in libraries. I haven't
> understood the data loss problem.
If "your" library crashes "my" application, I have no chance to do an
emergency save of the user's data, so presumably that's what Behdad is
getting at.
++
In the Java bindings, the Gtk-ERROR etc messages that result from these
g_return_if_fail() sanity tests are converted to unchecked Java
Exceptions. This helps enormously with debugging, but the developer can
_also_ do something like:
try {
Gtk.main();
} catch (Throwable t) {
data.emergencySave();
throw t;
}
or such, which means there is at least a chance of attempting to not
lose things.
Whether or not Java's unchecked exceptions is a good idea [or for that
matter whether C programs abruptly terminating on errors] is a topic for
another day :)
AfC
Sydney
--
Andrew Frederick Cowie
Operational Dynamics is an operations and engineering consultancy
focusing on IT strategy, organizational architecture, systems
review, and effective procedures for change management: enabling
successful deployment of mission critical information technology in
enterprises, worldwide.
http://www.operationaldynamics.com/
Sydney New York Toronto London
Attachment:
signature.asc
Description: This is a digitally signed message part