Because `globalValue` sux on swf9/10 and usually returns null, which
you can't dereference. André is right though, globalValue, or any of
the dereferences could result in a null that could cause an error in
the next step, and we don't want to see that error (the caller is
prepared to catch the error and work out longhand what the real issue
is).
This bug could be enlarged to say: Find all the places in the
debugger that expect an error and make sure that all the callees have
the appropriate pragma.
But the further mystery is that the try/catch is not inserted in
DHTML, only in swf9/10. So there is another mystery in the compiler...
On 2009-09-20, at 14:53, Henry Minsky wrote:
Why did a missing throwsErrors cause this? There are some probes of
various
places to see if the
symbol is found there, but aren't they all using the '[...]' access
mechanism, which doesn't throw an error if no
value is found?
On Sun, Sep 20, 2009 at 2:25 PM, P T Withington <[email protected]> wrote:
This is why you should always cc laszlo-dev:
http://jira.openlaszlo.org/jira/browse/LPP-8486
André already figured it out.
On 2009-09-18, at 20:29, P T Withington wrote:
Well, the debugger remembers errors and won't ever print the same one
again. So that could be why you don't see it again, or, something
is just
not initialized the first time around.
It will be a pain to debug, because fdb doesn't seem to intercept
errors
if they are/will be caught, and now that the debugger catches all
errors...
bleah!
I don't see it either, so I guess what you'd have to do is put a
#pragma
"throwsError=true" in (I think you could put it at the top-level
of the
class block) and try fdb-ing it?
On 2009-09-18, at 18:19, Henry Minsky wrote:
The first time I type in any 'simple' expression in a swf10
debugger, I
get
this error message. I don't see it happen again after
the first time.
lzx> Debug.window.x
ERROR @compiler/LzBootstrapDebugService.lzs#125: TypeError: Error
#1010:
A
term is undefined and has no properties.
71.25
lzx> Debug.window.x
71.25
lzx>
The line of code it is pointing to is evalSimpleExpr, but I don't
see why
it's generating an error.
--
Henry Minsky
Software Architect
[email protected]
--
Henry Minsky
Software Architect
[email protected]
