[PATCH] fix DBusError leaking libhal and libhal-storage

[David Zeuthen]

On Sat, 2006-07-22 at 21:51 +0200, Danny Kukawka wrote:
> On Saturday 22 July 2006 17:08, David Zeuthen wrote:
> > So, how about making
> [...]
> > complain to stderr if this macro is used but the error isn't set? I mean
> > we only should be using the macro if we know the error is set. With that
> > it looks good to me.
> 
> Only to understand correct: you want to print a message to the log if the 
> macro is used but the error is not set?
> 
> I don't see a problem to use the macro also if the error is not set, because 
> dbus_error_is_set() check if the error is set. This is IMO better than a 
> memory leak which cost much more than the check.

Yes, but want to fix up where we use it wrong - we should only be
freeing errors when we detect them.

> There are also two other possible solutions:
> 1.) return the error (via a additional pointer as parameter)

Would break API so probably not a good idea. ABI might be preserved (the
C ABI is preserved when just adding paramaters) but I don't like it.

> 2.) use NULL instead of a DBusError in function calls (currently there is no 
> information returned to the caller and we also not print any debug message 
> from the error)

I think the patch you wrote with a warning to stderr if the macro is
used incorrectly is the way to go.

Cheers,
David