IRC Log Viewer » #firebreath » 2012-08-14

IRC Nick Time (GMT-7) Message
Gordon_LN 07:08 I am not 100% sure when this started to happen (but I recently rebased onto 1.7). FF/Chrome all works fine. IE 32 bit any method which returns an array of items (int or string) is now showing up as undefined. IE64 register events fail (and the array issue).
FireBreathBot 07:08 Gordon_LN: 06 Aug 14:58Z <taxilian> tell Gordon_LN FYI if you make the width/height 0px some browsers will have issues; 1px by 1px is better
Gordon_LN 07:08 (the results return as undefined)
Libor 07:08 Hi all
I am logged into Confluence, but I can't log into forums
forum login says that I shall use the same name/pwd
When I try to log, I get this response:
If anyone from Forum Team could help me, it would be great
My username is "Libor P." if it can help
mkoch 09:08 hi
since my latest sys upgrade, the generator script doesn't run on Mac
error: There is no SDK with the name or path
I tried to specify the right path by -D CMAKE_OSX_ROOT="..."
but it seems to ignore it
(I can pas these arguments to the, right?)
hm. I symlinked the existing location to the one where the prep script looked for it. now it seems to work, at the end it displays the following error message: Manually-specified variables were not used by the project: CMAKE_OSX_ROOT
taxilian 09:08 mkoch: what version of cmake are you on?
mkoch 09:08 cmake version 2.8.7
taxilian 09:08 I think 2.8.8 is required to support then newest xcode; in fact, I believe there is a bug even there, and 2.8.9 (which may not be yet released) has the fix for that
mkoch 09:08 I will check, but for the moment it works so I'm happy with it
taxilian 09:08 glad to hear it =]
I also recommend you use the latest firebreath from master
mkoch 09:08 i use the latest dev usually :)
taxilian 09:08 good. update to get the changes I made over the weekend and let me know if the build times are any faster =]
Gordon_LN 09:08 is ActiveXBrowserHost::hasHTMLWindow still valid?
(I was using it to determine if the plugin was hosted in a web browser or a Win32 window) but now it always seems to return false (which is the cause of my IE issues)
taxilian 09:08 to be honest, I'm not sure. I need to head into the office; I'll be there and back on in 30-45 min at the outside. remind me then and I'll glance at it, might spot something useful
mkoch 09:08 taxilian: my biggest problem is that xcode eats up all the memory while I compile. I have 8gigs but it's not a problem for xcode to se it all and then freeze the whole system until the build is done
taxilian 09:08 mkoch: hopefully the change I made over the weekend (namely enabling pch) will help with that too
freezing the system is likely not memory but CPU; you can tell xcode not to use all of your cores
emicastro 09:08 taxilian: if I have two dll one for 32bit and one for 64bits... it's pluginconfig.cmake needs to be different? I mean the unique id like GUID and this kind of things?
taxilian 09:08 emicastro: no reason that I know of
emicastro 09:08 so, the same pluginconfig for the two dll should be right?
taxilian 09:08 as far as I know
32 bit COM and 64 bit COM can't collide with each other, AFaIK
emicastro 09:08 and I can have registered the two dll at the same time?
Gordon_LN 09:08 In the past I have used the same GUIDs for 32/64 on windows without issue
taxilian 09:08 they aren't in the same space, so certainly
32 bit and 64 bit don't register in the same part of the registry
so there should be no conflict
ok, I'm headed to the office
be back
taxilian 09:08 back
Gordon_LN 09:08 What I am seeing is that my some of my "host" references don't have any DOMDocument or HTMLWindow etc.
(I have just regenerated an empty project and I don't think I am missing any generated init code)
taxilian 10:08 Gordon_LN: which browser?
Gordon_LN 10:08 IE 32bit
taxilian 10:08 is the plugin hidden?
Gordon_LN 10:08 No - but I have three instances on the one page.
taxilian 10:08 shouldn't matter
all are completely visible?
no css being used to move them around, etc?
or change the scrolling on parent divs?
Gordon_LN 10:08 Yes visible except when I do my "hack" in chrome
FYI Chrome + FF working fine.
taxilian 10:08 95% of the time when I've seen an issue like that it's because of some CSS that people don't realize they're doing
Gordon_LN 10:08 Well I am not going to rule that out yet. FYI here is the specific issue I am looking at (I posted a few above).
Returning a simple type "int" works fine. (JS calling plugin method).
taxilian 10:08 how are you injecting it into the page?
Gordon_LN 10:08 Returning a list of int fails, because m_host is mostly NULL. Now based on your lat comment it could be that I am creating the control too early in my "widget" hierarchy and it hasn't actually become visible yet (or there is some initial resize css stuff going on).
Via innerHTML (or similar) assignment
taxilian 10:08 there is a big difference in many cases between innerHTML and "or similar"
how specifically are you injecting it?
jquery? dojo? somethign else?
Gordon_LN 10:08 dojo
But I have been messing with the sequence..
taxilian 10:08 can you tell me the line you're using?
I don't care about the sequence, for now
just how you're actually injecting it
Gordon_LN 10:08 Sure (it has changed a few times today mind):
taxilian 10:08 probably doesn't matter
Gordon_LN 10:08 this.plugin = domConstruct.create("object", { id: pluginID, type: "application/x-hpccsystemsgraphviewcontrol", width: "100%", height: "100%" }, this.graphContentPane.domNode);\
taxilian 10:08 yeah, that'll do it
don't do that
that will not work
it will produce bizarre problems
very much like the ones you're seeing, coincidentally
here is what happens there:
Gordon_LN 10:08 <g>
taxilian 10:08 1) it creates an object tag
2) it sets the id, type, etc
3) it adds the object to the dom
Gordon_LN 10:08 I think I see whats happening now - I am getting double constructs with the second one having invalid host info
taxilian 10:08 after 2) the activex control is created
after 3) it partially destructs the control and then rebuilds it
which tends to cause weirdness
insetad, create the html as a string, then use .innerHTML to put it into a div
Gordon_LN 10:08 ok, back to: this.graphContentPane.domNode.innerHTML = "<object"?
taxilian 10:08 yes
Gordon_LN 10:08 ok - I will report back (I got a fright as I had been working in FF/Linux and rebased on 1.7 then just today retested IE64 and it was all broken).
taxilian 10:08 ok
personally I still see no reason to support IE64, but that's me =]
Gordon_LN 10:08 BTW - Is there a good way to know when it is safe to register the events (I know you an't do it straight away)?
taxilian 10:08 sure. use the onload param
and do *not* create the object tag before inserting it into the dom
Gordon_LN 10:08 I _have_ to support IE64 as my control if for software developers who use eclipse to view the web page (and if they are using 64bit eclipse...)
taxilian 10:08 interesting
was not aware that 64 bit eclipse used 64 bit ie
this is why I ask =]
learn something new every day
Gordon_LN 10:08 (caught me by surprise also!)
Manduka 10:08 hi!
taxilian 10:08 good morning
Manduka 10:08 i´m trying to start developing on Firebreath, but i´m getting a lot of trouble installing "the enviromment" on windows
can you help me?
emicastro 10:08 windows 8 came with IE10 64bits
taxilian 10:08 that's difficult to say. "a lot of trouble" is rather unspecific; it could be that you're trying to use an apple ii or that aliens abducted your monitor, and there is no way for me to guess based on the information I have =]
Manduka 10:08 that´s a good answer :D
matutetandil 10:08 Hello
taxilian 10:08 hi
Manduka 10:08 first of all, i´m from spain, my english is not very well (sorry about that)
matutetandil 10:08 has someone use boost IPC in firebtreath??
taxilian 10:08 not as such, AFAIK
(to matutetandil)
matutetandil 10:08 ok...
beacuse I´m having some troubles with that...
taxilian 10:08 what kind of ipc?
matutetandil 10:08 the library is nt able to open a message queue to send messages
taxilian 10:08 ahh; you're the one who posted the issue
which browser is this?
matutetandil 10:08 yaps...
taxilian 10:08 Manduka: just do the best you can; we'll ask questions to clarify if we need to
matutetandil: first of all, have you verified with a normal windows app that you can open the queue?
Manduka 10:08 i trying to follow the videotutorial, and when i try to execute the prep2010.cmd to build my first plugin, i get this error: "Firebreath on windows requires ATL/MFC libs to installed. Download the Microsoft DDK and install the build enviroments in c:\WinDDk"
microsoft now have a new ddk, <- for windows 8
matutetandil 10:08 yes, I verified that, and work perfectly... I can send and receive messages
Gordon_LN 10:08 When you open your command prompt, open the the VS Tools command prompt
taxilian 10:08 Manduka: well, I haven't tried the new ddk; I don't know if it works, or how well
matutetandil: do you know how the message queue works? it may be that it conflicts in some way with how chrome is communicating between the plugin and the browser
Gordon_LN 10:08 Manduka - If you have the full VS2010 (which you probably have if its the one that came with Win8 RC) then just make sure you are runnint the prep from within "Visual Studio Command Prompt (2010)"
matutetandil 10:08 the queue is very simple to use and understand... I think the library is running inside a sandbox (that´s wired, beacause this kind of libs doesn´t)... but I don´t find any solution for this issue...
Manduka 10:08 \query Gordon_LN i have the express version
matutetandil 10:08 the queue exists and works... but the library can´t access it
that's wht
taxilian 10:08 matutetandil: there is no sandboxing on chrome's oop plugins that I'm aware of, unless they've changed something recently; however, they may be using their own queues that prevent it from working
I don't really know
have you tried it from firefox?
matutetandil 10:08 no, I don´t try in there... I need the lib to work in chrome... but I will make sone test in firefox
taxilian 10:08 see if it works; that would tell you something
matutetandil 10:08 ok... you are going to hear news soon...
thanks! and sorry for the issue report
taxilian 10:08 no worries
just nothing I can do about it =]
danzik17 10:08 taxilian: fyi, i ended up figuring out what was causing that error i was describing last night. needed to add this to my CMakeLists.txt: SET_SOURCE_FILES_PROPERTIES( modp_b64.c PROPERTIES LANGUAGE CXX )
fb latest does seem to build a bit faster fyi
taxilian 10:08 wait, what?? you needed to build the .c file as a c++ file?
that's more than a little bizarre
danzik17: re: speed. cool; I updated the pch to be project-wide instead of applying to each file, which should let visual studio's optimizations work a bit better
danzik17 10:08 yea, compiling it as C++ works, no idea why really. i took a diff of the vcxcproj immediately after running the prep scripts and then after including the .c source manually
taxilian 10:08 good thinking
danzik17 10:08 the difference ended up being that when included manually, it was being compiled as c++
taxilian 11:08 that is really really odd
Gordon_LN 11:08 c/c++ will produce different name mangled functions which would cause issues at linkage time.
Had the headers been decorated correctly it would not be an issue, but a lot of the time they are not.
taxilian 11:08 yeah, that could be it
would you mind dropping a note in tips and tricks about that?
Gordon_LN 11:08 Sure - but its not really a firebreath issue....
taxilian 11:08 yeah, but it's useful for firebreath users to know
Gordon_LN 11:08 (I have a few .c files, which are working without any magic - trying to see what I did differently)
Whn you include the files you could put them in:
#ifdef __cplusplusextern "C" {#endif
#ifdef __cplusplus
extern "C" {
} #endif
Or ideally that should be in the c headers them selves...
danzik17 11:08 Gordon_LN: that actually is there, just looked.
taxilian 11:08 wow… is amackera actually here?
pitb00l 12:08 Hello everybody, I got a question which possible is also a sugestion - I downloaded the FireBreath lib and was amazed about the lib itself, but I didn`t find it so easy to work with as stated - what I need is just a plugin exporting one single function receiving no parameters and returning just one string - thats it. yet it seems to complicated to figure it out by looking at the examples.
matutetandil 12:08 Hi taxilian...
I test my issue in firefox...
same thing :(
I cannot open the queue
jshanab 12:08 pitb00l. FB is a framework not a library. You have almost described the default plugin if you run the fbgen script and compile. Did you start there?
oh, duh he left before anyone could answer
Gordon_LN 12:08 Well IE32 is working fine again.
taxilian 12:08 I'm back from lunch
wow, lots of newcomers today
Gordon_LN 12:08 IE64 is still getting double constructs for each instance, what is possibly interesting is that the origins are "jscrip9.dll" and "mshtml.dll" for each instance.
jshanab 12:08 IE32? Holy cow, I can't even get IE8 to work correctly ;-)
taxilian 12:08 lol
Gordon_LN 12:08 <g>
taxilian 12:08 Gordon is trying to be proactive
he wants to stay ahead of the curve. he got ie30 and 31 working yesterday
Gordon_LN 12:08 Well with FF that could be somtime in the next month or so<g>
taxilian 12:08 lol
matutetandil 12:08 Hi taxilian... do you read the results of IPC in firefox???
don´t work :(:(
taxilian 12:08 yeah, I saw. definitely something weird going on
but I don't know what it would be; perhaps because you're a child process of another process?
matutetandil 12:08 perhaps... But i'm not sure... It has to work...
taxilian 12:08 probably does, but maybe you need to set something up, tweak some permissions, something similar
matutetandil 12:08 I don't know what to do... it's very wired...
taxilian 12:08 sorry I can't help
matutetandil 12:08 don't worry
be happy
taxilian 13:08 lol
matutetandil 13:08 ;)
If I make it wok, I´ll send a how-to :):):)
taxilian 13:08 sounds good
emicastro 13:08 have you ever register two dll (one for 32 an other for 64 bits) in the same machine that the two of them handled the same mime type?
taxilian 13:08 personally? no, but there is no reason it wouldn't work
I know where in the registry both go and they aren't related
also I'm pretty sure a couple of people in here have
emicastro 13:08 could you say me in wich part of the registry I have to look in order to see if is everything ok ?
taxilian 13:08 you can find it in the .rgs file
emicastro 13:08 I know that I'm doing wrong, cause I'm registering the dll with regsvr32 command.... (but this is something that I did not decide to do :S)
taxilian 13:08 and you're using the 64 bit regsvr32 to register the 64 bit one, right?
hmm; an interesting point, it should work for IE but the last one you register may override the npapi plugin path
emicastro 13:08 ohh jesusssss!!! kill me nowww!!! I will check if the guy that do the installer is using the regsvr64 !
taxilian 13:08 I understand your sentiment, but I would really prefer if you could keep your language free of such exclamations in the future
emicastro 13:08 yeah, the problem that I have is that in some point, one dll is overriding the other
taxilian 13:08 I actually find swearing in that particular manner to be extremely offensive, though I understand that it was not intended that way
Gordon_LN 13:08 You use regsvr32 to register 64bit dll
emicastro 13:08 No, sorry... It is not my intended... I really sorry... I do not know the consecuences of use this kind of language in other countris/languages different than mine... I really sorry, I won't use it anymore
taxilian 13:08 emicastro: don't worry, I said I find it offensive, not that I was offended at you =] something can be offensive, but we can still choose to take offense
in fact, it's not that uncommon to be used in this country either; it's more of a personal and religious thing. don't feel bad, just please don't use that name again in here =]
emicastro: yeah, I think it's still called regsvr32, but there are 64 and 32 bit versions; dunno if it handles the difference automatically or not
(I meant we can still choose not to take offense in my comment earlier)
emicastro 13:08 You're right... ok, I got it! I'll try with the regsvr to 64 bits! I hope this solve the problem!
taxilian 13:08 let me know how it goes =]
emicastro 13:08 There is not a regsvr64... I've read that by default the regsvr32 run in 64bit mode... and if the dll is 32bit regsvr32 will call the regsvr32 from syswow64 directory...
mmm but I don't believe that :)
taxilian 13:08 that sounds consistent with what others seem to be saying
what is and what is not working?
emicastro 13:08 My installer run regsvr32 for 32bit dll and then regsvr32 for a 64bit dll.... In a 64bits windows 8, thats came with a 64bit IE10 installed, the plugin is working fine...
but if you install a normal downloaded version of chrome or firefox (that is 32 bits) they does not see the plugin registered
the plugin is not in about:plugins list...
taxilian 13:08 ahh
register the 64 bit one first, then the 32 bit one
should fix the problem
I thought you were talking about IE
emicastro 13:08 I'll try that way...
I hope this work fine!!!
taxilian 13:08 I don't actually know how to register a 64 bit plugin for firefox or chrome
interesting question
emicastro 13:08 if you use a 64 bits Chrome or Firefox... with the regsvr32 is ok
taxilian 13:08 there is probably a way to specify both a 32 and 64 bit plugin
but right now they override each other
emicastro 13:08 but the problem is, that for download a 64bit version, you have to do it explicity
taxilian 13:08 right
emicastro 13:08 and if you go to chrome download webpage, they will give you the 32 bit version
and there is when the problem start! :)
you have a win 64 bit version, with a IE native 64 bit version.... and then you install a firefox or chrome 32 bit version
And the registry look like cann't determine wich dll has to use...
I'll try registering the 64 bit version first
taxilian 13:08 there is very rarely a real reason to use the 64 bit version of firefox or chrome on windows
emicastro 13:08 yeah... but requirement team does not anything about this! hahaha
*does not know
taxilian 13:08 yeah
Hmm. okay, I now know how to fix the issue; we need to make it explicitly put the npapi plugin in the Wow6432Node part of the registry
add a jira tag, I'll try to get to it
need to figure out how best to do it :-/
emicastro 13:08 ok, I'll create the jira issue
taxilian 13:08 or of course you're welcome to do it yourself and send a pull request =]
emicastro 13:08 It's seem that switching the registration order (64bits first) is working...
taxilian 13:08 right; that's because the 64 bit was overriding the 32 bit registration for npapi
but it wont' work on 64 bit npapi as it is now
emicastro 13:08 I have to generate a new intaller and make some test... I'll tell you the results. Ok, I'll look for my time to do the fix, if you help me in working on it of course :)
taxilian 13:08 well, I'm not really known for being helpful, but I'll try to make an exception
emicastro 13:08 so if I'm install a 64bits Chrome or Firefox, this switching fix won't work?
taxilian 13:08 correct
emicastro 13:08 haahah this attempt to my first goal, that was support Chrome and Firefox 64bit version :P
taxilian 13:08 lol
emicastro 13:08 this look like a Möbius strip! lol
taxilian 17:08 take care, y'all
dougma 18:08 mmm... firefox and chrome exist in 64bit versions?! I did not know that....
johannes 19:08 taxilian_away: I don't know how mauch breaking of APIs you like, I was careful while plaing with my file upload things: