Bug#551251: libjavascript-perl: FTBFS: failed test

[Mike Hommey]

On Sat, Oct 24, 2009 at 04:01:13PM +0200, Mike Hommey wrote:
> On Sat, Oct 24, 2009 at 02:48:37PM +0200, Mike Hommey wrote:
> > On Fri, Oct 16, 2009 at 10:31:23PM +0200, gregor herrmann wrote:
> > > From t/21-branch-handler.t:
> > > 
> > > 22 $cx1->set_branch_handler(\&branch_handler);
> > > 23 $cx1->eval("for(i = 0; i < 10; i++) {}");
> > > 24 is($called, 10);
> > > 
> > > 
> > > Hm, I guess it's not that easy :)
> > 
> > It's actually almost that easy. You should also change the
> > PJS_branch_handler definition to drop its second argument.
> > 
> > As for the test, with the API change, the assumption that the "branch"
> > handler is going to be called at each for iteration becomes wrong, which
> > is why the test fails.
> > 
> > Unfortunately, while I have been able to verify that indeed the callback
> > is properly called by the JS engine if I hack the JSContext members to
> > make it happen, I couldn't get it to trigger under normal conditions
> > with an eval()ed script :(
> 
> After further investigation, it appears it is not triggered at all, at
> the application itself has to trigger it with
> JS_TriggerOperationCallback.
> 
> See
> http://groups.google.com/group/mozilla.dev.tech.js-engine/browse_thread/thread/a4d1fe147761aacb/e61d2592faf4ef72?lnk=gst&q=js_setoperationcallback#e61d2592faf4ef72
> for reference.

As there is no reverse dependencies to libjavascript-perl apparently, I
would see no risk in changing the API.

As the branch handler is not working the same way anymore, I'd say it's
safe to simply remove it, until someone (upstream?) comes up with a
proper perl API to do the JS_{Set|Trigger}OperationCallback magic.

Mike