Re: [g-a-devel]at-spi patch ...

[Michael Meeks <michael ximian com>]

On Fri, 2002-08-16 at 13:14, Bill Haneman wrote:
>  I take it that the change to two bitfields (on_loan,
> ref_count) from is considered bincompatible, since the ref_count field
> was not exposed?  

	If it's not in an installed header, and you rely on it - then you can
expect to be broken - yes. Hence it's not similarly not reasonable to
do:

	extern char * cspi_exception_get_text (void);

	in your app, and then call that. You get to get broken later ;-)

> In case you were wondering about my use of local 'retval' variables in
> some places (which you removed), I found them useful when using gdb.  No
> matter.

	gdb seems to suck more and more; hmm.

> The only actual issue I have is with the commenting-out of the key
> listener tests in test-simple.  As you might guess, they "work for me",
> thus I'd rather get to the bottom of why they are failing for you than
> comment out an important regression test.

	Well - I have a fairly stock RH 7.3 install, with sawfish 1, running
the test on an xterm, and I couldn't get it to pass once in 3 tries so
...

> Please commit, but understand that I intend to re-enable keylistener
> testing by default, and try to squash the bug you are seeing.

	Committed,

		Michael.

-- 
 mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot