IRC Log Viewer » #firebreath » 2012-06-20

IRC Nick Time (GMT-7) Message
dougma 07:06 re FB-182: i can't reproduce.
i shall say as much in the jira. :)
taxilian 07:06 I just did =] not sure why FBBot didn't pick it up, tho...
thanks for looking into it
also good catch on the refcount in activex thing the other day =]
dougma 07:06 oops... double comment. :)
no probs. your comments contradicting the code made it easy
taxilian 07:06 hehe
dougma 07:06 and that does indeed fix fbtestplugin too.
taxilian 07:06 excellent
dougma 07:06 will have a look at fb-177 now
taxilian 07:06 FIREBREATH-177
… what is with my bot today?
dougma 07:06 funny. :)
taxilian 07:06 *sigh*. so hard to get good help these days
actually, it's probably not working bceause I moved jira to another server...
dougma 08:06 anyone ever successfully used atl tracing in firebreath?
well, it works... now to make sense of it all. :)
taxilian 09:06 dougma: I've never even figured out how to use atl tracing at all
dougma 09:06 there's this ATL/MFC trace tool program as part of visual studio
taxilian 09:06 ahh
dougma 09:06 you use it to enable the tracing categories/levels
and then the messages go via OutputDebugString (so to debugger, or debugview)
the right tool for this particular job
though i've never used it previously!
taxilian 09:06 ahh
makes sense
what are you tracking down?
177 still?
dougma 09:06 yeah
taxilian 09:06 you were able to repro?
dougma 09:06 cause i'm some kind of masochist
taxilian 09:06 well I sure appreciate it; I've been very slim on time lately, and for some reason ATL issues aren't usually the top of my list of things I want to spend time on….
and you're probably the only one who is frequently around besides me who is technically competent to do any significant work on that part of the code
dougma 09:06 no worries
taxilian 09:06 so is InPlaceActivate being called?
dougma 09:06 yes, the control is coming up windowless
so, just moving the init of m_bWndLess from SetObjectRects into InPlaceActivate
taxilian 09:06 huh; SetObjectRects isn't being called?
dougma 09:06 it is, but m_bWndLess has been set to true before that happens
so atl doesn't create the window
taxilian 09:06 I think I need to go look at that file again =]
dougma 09:06 i'm not quite there
zeroing in.
taxilian 09:06 hmm. can you think of any reason why m_bWndLess should not be set in setReady when we set m_bWindowOnly?
similarly, why would setting m_bWindowOnly not cause this to work correctly?
Have you verified that either InitNew or Load are being called?
dougma 09:06 got it.
taxilian 09:06 ?
dougma 09:06 the atl code is initialising m_bWndLess
taxilian 09:06 right
but it should generally do so taking inot account m_bWindowOnly
dougma 09:06 unfortunately not.
atlctl.h line 963
oh yeah, it checks m_bWindowOnly above that
taxilian 09:06 right
so when this is called, that means that m_bWindowOnly hasn't been set
which means that somehow neither InitNew or Load was called
which technically shouldn't be possible, but would match the symptoms
dougma 09:06 when does setReady get called?
i'll bp it. :)
taxilian 09:06 since those two are the methods relating to IPropertyBag which is how the params are passed
setReady is called from InitNew and Load
InitNew should be called if there are no params
dougma 09:06 that matches, indeed.
taxilian 09:06 so if that's the case, then setReady is never called
but one of those two should *always* be called, and in the past it was
so what changed?
dougma 09:06 InitNew is from IPropertyBag too?
taxilian 09:06 yes
dougma 09:06 so, if there's no properties, it doesn't get called i suppose.
taxilian 09:06 sorry, IPersistPropertyBag
the way that it has always worked before is if there are params, Load is called; if not, InitNew is called
dougma 09:06 the all new ie9?
taxilian 09:06 this is certainly a possibility
dougma 09:06 i've never used params
and always been windowless. :)
taxilian 09:06 heh
dougma 09:06 well... time for me to go.
taxilian 09:06 hmm . okay. have a good night
PauloMarcio 12:06 taxilian: I found something about the bug with
taxilian 12:06 yeah?
PauloMarcio 12:06 if NPJavascriptObject::HasProperty("ownerDocument") return false the bug does not occur
taxilian 12:06 then call setReserved("ownerDocument") in your root JSAPI object's constructor
PauloMarcio 12:06 hum, I'll give a try... tnx
it worked!! Tnx a lot!
do you consider change that in JSAPIAuto::init( ) ?
taxilian 13:06 I would probably accept a pull request that had that in it, yes
PauloMarcio 13:06 hum, nice... and how do I do that?
I supose it's here
taxilian 13:06 no, on github
EL45 13:06 Hey taxilian just to let you know, I'm receiving Django 404 errors when trying to view the forum.
taxilian 13:06 grr
I hate that forum
EL45 13:06 Sorry :(
EL45 13:06 So i've written a shim to go between my plugin and a .Net assembly, everything is working outside of firebreath(ie from an unmanaged c++ console app). However, when I add my code to my plugin, my shim can't find my .NET assembly ("Could not load file or assembly 'ClassLibrary1')" All three of the dll's are in the same directory. Any suggestions off the top of your head?
taxilian 13:06 SetDllDirectory?
remember that the DLL in question is not the executable
EL45 13:06 Ok ill look into that. My shim calls "#using "ClassLibrary1.dll" rather than LoadLibrary.
EL45 13:06 I noticed my test app has my shim listed in both references and linker input, while my Firebreath plugin just shows the shim in the linker input.
However I can't manually add a reference to test if thats the problem.
taxilian 13:06 references is a .net concept