IRC Log Viewer » #firebreath » 2011-01-21

IRC Nick Time (GMT-7) Message
FB_GitHubBot 00:01 FireBreath: master Richard Bateman * d23e36d (26 files in 6 dirs): IE reference count fixed, windowless support ... -
FB_GitHubBot 00:01 FireBreath: master Richard Bateman * eda0bc8 (4 files in 1 dirs): Fixed a few missed changes for mac build -
Ed____ 06:01 hey. For anyone who was following yesterday..
I have got to the point where I can build my shared libraries as part of the plugin.. :D
In the main CMakeLists.txt, I did add_subdirectory(path ${FireBreath_BINARY_DIR})
target_link_libraries(${PRONAME} libararyname)
Then just created a CMakeList.txt in the library directory, and it has ${FireBreath_BINARY_DIR}ut it all together, and seems to be building.
I havent added the shared libraries to the WiX install yet, but the hardest part was putting the above together..
iaincollins 06:01 oh cool, thanks Ed____
Ed____ 07:01 thanks for the pointers..
Anbody a glob guru, while I am [email protected]
I wanted to create file list using a glob pattern, that excluded a single specific filename.
Done it with list (REMOVE_ITEM), but just wondered if anyone knew some magic..
[dudu] 08:01 hi
taxilian 08:01 g'morning all
[dudu] 08:01 good morning to you too (its evening here, btw :-))
taxilian 08:01 hehe. yeah, I figured =]
thanks for posting that info; that helps me understand better what may be happening
we'll see if it's enough to figure out why it's happening
[dudu] 08:01 that's good to hear, because for me it's all foggy what's going on
taxilian 08:01 well, this much I know:
it's crashes when GTK sends an event
because somehow the event is coming back on a bad ptr
so if you look at PluginWindowX11.cpp line 167, that's where the crash originates, basically
it's getting there without a valid pointer
maybe more specifically you could say it originates on line 148
based on what you say, it sounds like user_data is not valid
do you know much about how g_signal_connect works?
[dudu] 08:01 nope, i'm all new to the whole gtk/gdk/x11 stack
taxilian 08:01 me too, unfortunately
I'm wondering if there is some teardown that we should know about
kalev: you around? any ideas?
this is hte only issue holding up 1.4 beta 2
[dudu] 09:01 i've a gdb session in front of me, i dont know if that helps, but i can pastebin you a print of the user_data from there
taxilian 09:01 I assume that the ptr is not null?
can you pastebin me what it looks like when it is cast to a PluginWindowX11?
I tried to reproduce this last night and again this morning and it isn't happening on my VM
[dudu] 09:01 sure,
taxilian 09:01 hmm. interesting. the odd bit is that there isn't quite enough information to know if the ptr is still valid or not :-/
were you refreshing the page or anything like that during this?
when it crashed?
[dudu] 09:01 just clicking on the set event handler, clicking on the plugin surface as far as i can tell
taxilian 09:01 hmm
are you using this from git? so that it'll be easy to update if I post a fix to try?
[dudu] 09:01 yes, it from github
taxilian 09:01 [dudu]: could you switch to branch master (if you aren't already there) and try the latest?
FB_GitHubBot 09:01 FireBreath: master Richard Bateman * c760ff5 (3 files in 2 dirs): Attempted fix for mutex issue on X11 -
[dudu] 09:01 ok, pulled it down, compiling
kylehuff 09:01 taxilian: I just did a pull, rebuilt my build dir using and it failed with this:
Ed____ 09:01 Hi again, got a WiX related question today..
So, I am adding my built Dlls to the WXS, and I need to supply a Source path..
taxilian 09:01 kylehuff: you're certain you got the latest? it works for me… that sounds like you somehow updated the .cpp and not the .h
Ed____ 09:01 The source path needs to be configuration dependant..
I have tried ${CMAKE_BUILD_TYPE} and ${ConfirgurationName}, but these are both empty in the generated WXS..
kylehuff 09:01 taxilian: I did a git pull using the master branch;
taxilian 09:01 Ed___: look at the bottom of Win/projectDef.cmake
you'll notice a path/filename for np${PLUGIN_NAME}.dll
that's the main plugin DLL
you should be able to tweak that to get what you need
kylehuff 09:01 (btw; when you get time - I see there are other questions being asked here, I am in no hurry, feel free to ignore me until it is convenient..)
[dudu] 09:01 well, now it doesn't seem to crash, but when i click on the "fire event" it seem to fire it like 10 times (get a bunch of alert with "onfired(): test stuff" and one "onfired() from load(): test stuff" at the end
Ed____ 09:01 Ah, ok. I add the Dlls there? or make use of ${CMAKE_CFG_INTDIR}?
taxilian 09:01 Ed___: make use of ${CMAKE_CFG_INTDIR}
[dudu]: even if you only click attach once?
kylehuff: something strange, then; it works for [dudu] and I, aparently, and from this you can see that m_handler_id is there in the source:
[dudu] 09:01 yup, but that could be the artifact of gdb, without running in gdb seems to be working normaly
taxilian 09:01 and no crash?
Ed____ 09:01 Aha. Ok, Looks like I can use the CMAKE_CFG_INTDIR and that puts in the right path. Cheers Taxillian..
taxilian 09:01 hmm. very interesting. it must be detaching and reattaching the window
Ed___: glad I could help
brb, my wife just made me breakfast, so I better go eat it. shouldn't be gone long
[dudu] 09:01 nope, it didn't crashed until now (-:
Ed____ 09:01 Taxiliian: Cheers. Eggs and bacon? Yum.. :D
[dudu] 09:01 hmm, maybe i mislead you, it seems to add one more "test stuff" every time i click on the "set event handler" link above the "Fire event"
i guess that's what suppose to happen according to the js in the page
jshanab_wcw 09:01 I am having an urgent non-Firebreath issue (the reason I am moving to FireBreath) but I am hopeing someone can offer a hint. I have a cab file containg a plugin that installs and works fine from our domain. Then when a copy of the site is made to another server in a lan or even another domain, it never finishes the install and IE goes into an infinite loop of installing.
taxilian 09:01 [dudu]: yep, that's normal
so it seems to be stable now?
[dudu] 09:01 yup, seem to be unbreakable (-:
taxilian 09:01 excelent
[dudu] 09:01 thanks a bunch
kylehuff 09:01 hmm.. I am at a loss as to what my problem is; this is the entire bash buffer of exactly what I did after I saw master had been updated.
taxilian 09:01 thank you, I've been worried about that one
jshanab_wcw 09:01 BTW taxilian, I got a pretty good plugin going now, thanks
taxilian 09:01 kylehuff: run git submodule update
jshanab_wcw: awesome =] what version are you using?
jshanab_wcw 09:01 I am still on the earlier version, i think 1.2. I had to stop to try and solve this emergancy issue with the existing plugin that no one can solve
taxilian 09:01 ahh. yeah, you definitely will want to upgrade
hopefully you're on 1.3.2 and not on 1.2
but still
kylehuff 09:01 no change after submodule update
jshanab_wcw 09:01 I do have a crash on plugin exit problem, that is my next task
taxilian 09:01 kylehuff: hang on, that might be my fault, actually...
you're using FB_GUI_DISABLED
kylehuff 09:01 if I do a `git checkout firebreath-1.4` and re-create my build dir, it appears to work fine..
taxilian: Yes, I am
taxilian 09:01 I overlooked that
my badf
I'll fix it =]
glad you tested
kylehuff 09:01 no problem; I was only testing it so things like this could be found..
taxilian 09:01 I'm sure glad you did
kylehuff 09:01 next time I have a problem, I will remind you I'm using FB_GUI_DISABLED..
[dudu] 09:01 should i post an update to the newsletter thread with something like a "fixed"?
taxilian 09:01 yeah, but hang on until I have it in the 1.4 branch
[dudu] 09:01 well, it's not urgent for me, i'm just building little prototypes
taxilian 09:01 [dudu]: I appreciate your help tracking this down
only one other person had reported it
and I haven't been able to reproduce it, which makes it very difficult to fix =]
[dudu] 09:01 no problem, i'm glad i could help
FB_GitHubBot 09:01 FireBreath: master Richard Bateman * e854bd0 (2 files in 1 dirs): Fixed FB_GUI_DISABLED linux error from last commit -
taxilian 09:01 kylehuff: let me know if that fixes it
kylehuff 09:01 ok, will do
taxilian 09:01 I think we're ready for FireBreath 1.4 beta 2, as soon as kylehuff confirms this working
kylehuff 09:01 already further than before..
taxilian: it works now
taxilian 09:01 excelent
kylehuff 09:01 I made both the plugin, and all the tests, just to be sure.
taxilian 10:01 well, I'm going to run an extra build on all supported platforms just to make sure, but then I think i'll release 1.4 beta 2
[dudu] 10:01 good bye
kylehuff 10:01 I am about to do a test build on Linux 64
taxilian 10:01 cool
kylehuff 10:01 my connection to home is slow as hell today though..
taxilian 10:01 no negative sounding "hmms" allowed here, please
it works for me, therefore it is perfect
paztulio 10:01 How do I throw an Exception into browser JS?
kylehuff 10:01 oh, wait, nevermind.. I need to update my plugin
taxilian 10:01 paztulio: throw FB::script_error("message")
paztulio 10:01 Thanks.
taxilian 10:01 np
kylehuff 10:01 okay, all tests passed on x86_64 also
taxilian 10:01 excelent
FB_GitHubBot 10:01 FireBreath: master Richard Bateman * 0da20ac (0 files in 0 dirs): Merge branch 'firebreath-1.4' -
FireBreath: firebreath-1.4 Richard Bateman * d23e36d (26 files in 6 dirs): IE reference count fixed, windowless support ...
FireBreath: firebreath-1.4 Richard Bateman * eda0bc8 (4 files in 1 dirs): Fixed a few missed changes for mac build
FireBreath: firebreath-1.4 Richard Bateman * c760ff5 (3 files in 2 dirs): Attempted fix for mutex issue on X11
FireBreath: firebreath-1.4 Richard Bateman * e854bd0 (2 files in 1 dirs): Fixed FB_GUI_DISABLED linux error from last commit
FireBreath: firebreath-1.4 Richard Bateman * 0da20ac (0 files in 0 dirs): Merge branch 'firebreath-1.4'
FireBreath: firebreath-1.4 commits a34022f...0da20ac -
paztulio 10:01 shouldn't it show "Error: message" instead of "Error: Error calling method on NPObject." inside the browser?
taxilian 10:01 that's a bug in chrome
it has been filed with the chromium team, but not fixed yet
paztulio 10:01 They have more important stuff to do... Chrome / Webkit has a massive memory leak in the DOM... :(
taxilian 10:01 lol
paztulio 10:01 whenever a node is removed or changed, it leaks... o_o
taxilian 10:01 heh. yeah, kinda stinks that there is nothign we can do about the exception thing, though
paztulio 10:01 I stared the bug. Hopefully, it gets some attention soon.
taxilian 10:01 worth a try
taxilian 12:01 lol. so I posted firebreath-1.4 beta 2 and then found something almost immediately that I was supposed to do before that
ahh, well
it'll be in the next release
which I hope will be RC1
it's just the invalidatewindow call on IE for windowless plugins
taxilian 13:01 for anyone interested in setting up gitolite: new FireBreath related blog post: : Git and gitolite nirvana
grrr. not firebreath related. copied the wrong line :-P
scJohn 13:01 dang, i need to give someone else access to a git resource. looks like gitosis is what i need. :)
taxilian 13:01 gitolite
not gitosis
gitolite is much nicer
I can give you more details if you'd like
nirvdrum 13:01 I'll have to check that out. I currently use gitauth:
FB_GitHubBot 13:01 FireBreath: firebreath-1.4 Richard Bateman * db4aa0c (10 files in 3 dirs): Added callback for invalidatewindow in IE windowless ... -
taxilian 13:01 gitolite is way more awsomer
I'm sure of it
now I'll go see what gitauth is
ahh. not bad
gitolite nominally uses a git repo for configuration, though if you set it up like I did you don't need to use it a whole lot
mostly just for adding new users
though you could write your own (ruby?) frontend for that if you wanted
does gitauth support wildcard repos?
i.e. you can just push to [email protected]:username/new_repo.git?
looks like there are advantages to both; I think gitolite is a little more what I need, though. particularly like the integration with git-daemon
scJohn 13:01 do they require access to a port?
taxilian 13:01 tcp port? git-daemon uses a port. git://...
scJohn 13:01 yea, that was my question
taxilian 13:01 mostly gitolite works like gitauth, though; usually you want to use a key
scJohn 13:01 so you have to open up something other than ssh.
taxilian 13:01 but for public read acecss daemon works
scJohn 13:01 ok
that was my basic question
taxilian 13:01 there is also a "mob" access that may go over ssh
haven't looked into yet
huh. cool.
scJohn 13:01 i'll probably have to setup a user short term as ssh is the only port i reall have access to.
taxilian 13:01 allows anonymous ssh access
scJohn 13:01 wow
taxilian 13:01 so basically "if you're not a key I know, you are mob"
and then you control what mob has access to
that's actually pretty cool
slightly dangerous, but if used properly… quite cool
scJohn 13:01 slightly
FB_GitHubBot 13:01 FireBreath: master Richard Bateman * db4aa0c (10 files in 3 dirs): Added callback for invalidatewindow in IE windowless ... -
scJohn 13:01 did i read where the 1.4 branch has been updated?
taxilian 13:01 yes, 1.4 is now at beta 2
scJohn 13:01 k, thanks.
taxilian 13:01 and master is safe again, btw
in fact I've combined them to be the same thing again
it was too confusing and I'm not really doing 1.5 features yet
headed to lunch now. be back in a bit
scJohn 13:01 cya
FB_GitHubBot 16:01 FireBreath: master Richard Bateman * bc48a84 (1 files in 1 dirs): Removed unneeded duplicate line -
FireBreath: firebreath-1.4 Richard Bateman * bc48a84 (1 files in 1 dirs): Removed unneeded duplicate line -
FireBreath: firebreath-1.4 Richard Bateman * f18a2c2 (2 files in 1 dirs): Fixed bug in carbon event handling -
FireBreath: master Richard Bateman * f18a2c2 (2 files in 1 dirs): Fixed bug in carbon event handling -
scJohn 16:01 taxilian, do I need to re-run the prep scripts after updating from git?
taxilian 16:01 depends
if any files have changed, then yes
if not then no
cmake files
scJohn 17:01 ha
taxilian 17:01 or files were added or removed (or moved)
scJohn 17:01 i can just re-run every time.
taxilian 17:01 not a bad plan
scJohn 17:01 has been my sop so far, just making sure it is necessary.
anything i should be looking at in the latest beta? anything you specifically need tested?
taxilian 17:01 just general stability
unless you need windowless support
scJohn 17:01 k, that i can do.. actually i do, but i don't really care about a 10x10 window sitting there (I assume that is what you mean by windowless)
taxilian 17:01 well, not really
but you can reduce the possibility that it will cause you problems by making it windowless
windowless doesn't mean that it doesn't draw
it means it doesn't get its own window
scJohn 17:01 oh
taxilian 17:01 it draws in the browser window
scJohn 17:01 i'll be glad to test
if you point me in the right direction
taxilian 17:01 so if you return true from isWindowless it will not create an extra window that (even at 10x10) will float over the top of anything on the DOM page, which could be nice
scJohn 17:01 ok, i'll try later when i get back on.
taxilian 17:01 you don't use any GUI, right?
scJohn 17:01 no, i don't use anything on the gui
taxilian 17:01 then do set(FB_GUI_DISABLED 1) in your PluginConfig.cmake
and let me know if that causes any problems
scJohn 17:01 will do
the cmake is all i need, i don't need to worry about isWindowless?
taxilian 17:01 no
that will force it to be windowless
scJohn 17:01 well, chrome was happy. ie not so much...
taxilian 17:01 symptoms?
scJohn 17:01 well
it may be un-related
taxilian 17:01 what did it do?
scJohn 17:01 it is complaining about a random javascript line..
taxilian 17:01 huh. yeah, that doesnt' sound like a plugin problem
though you never know
scJohn 17:01 but that js file has not changed... strange. i will debug and let you know. like you said probably not related
taxilian 17:01 hmm. there is definitely a problem with FB_GUI_DISABLED on windows, though
but it shouldn't cause what you describe
scJohn 17:01 i reverted to a previous version and it was still not happy
taxilian 17:01 probably your javascript, then
IE is really finicky about js
scJohn 17:01 thanks
oh well.. maybe later cya
taxilian 17:01 I'll have FB_GUI_DISABLED fixed shortly
try it later =]
FB_GitHubBot 17:01 FireBreath: firebreath-1.4 Richard Bateman * aa985a2 (5 files in 2 dirs): Fixed FB_GUI_DISABLED on windows/IE -
FB_GitHubBot 18:01 FireBreath: firebreath-1.4 Richard Bateman * b2c5940 (2 files in 1 dirs): Fixed bug in security zones w/ JSAPIProxy -
FireBreath: master Richard Bateman * aa985a2 (5 files in 2 dirs): Fixed FB_GUI_DISABLED on windows/IE
FireBreath: master Richard Bateman * b2c5940 (2 files in 1 dirs): Fixed bug in security zones w/ JSAPIProxy
FireBreath: master commits f18a2c2...b2c5940 -
mgutz 21:01 Cannot compile firebreath. Lots of errors around unit32_t and int32_t
undeclared identifier
Source from github master. Any advise?
Win7 x64, VS2008 Debug Win32 targt
taxilian 22:01 mgutz: Hang on
mgutz 22:01 k
taxilian 22:01 could you pastebin the errors?
mgutz 22:01 k
taxilian 22:01 not sure why, but some computers have errors that don't show up on others
of the same configuration
dealing with those types
so I don't always catch them
mgutz 22:01
taxilian 22:01 hang on
FB_GitHubBot 22:01 FireBreath: firebreath-1.4 Richard Bateman * 167fe36 (1 files in 1 dirs): Attempt to fix uintN_t types for all browsers -
taxilian 22:01 try that
FB_GitHubBot 22:01 FireBreath: master Richard Bateman * 167fe36 (1 files in 1 dirs): Attempt to fix uintN_t types for all browsers -
taxilian 22:01 let me know if it fixes it
mgutz 22:01 success, 0 errors THANKS
taxilian 22:01 no problem. thank you for letting me know
you just getting started with FireBreath?
mgutz 22:01 yes, want to create an xmpp plugin for a chrome based web app
taxilian 22:01 cool
commercial or hobby?
mgutz 22:01 startup
taxilian 22:01 sounds neat. where you located?
mgutz 22:01 San Diego, CA, u?
taxilian 22:01 Utah
There are usually people here during the day, but you almost missed me
my first response was when I popped in on my cell phone to see if anything new had happened =]
so if you drop in and we're not here, dont' despair; just try back
mgutz 22:01 thank you mcuh :)
taxilian 22:01 I look forward to hearing how your stuff goes
mgutz 22:01 thanks, will try to make a small donation. this could make the plugin really easy
taxilian 22:01 money is good, but please just contribute in whatever way(s) you can. the help docs on the wiki need some love as well
mgutz 22:01 ok, will do
taxilian 22:01 there is plenty to do =] and if everyone contributes just a little it goes a long ways
but then, working at a startup I imagine you understand that =]
mgutz 22:01 ;)
taxilian 22:01 when you say Chrome; you talking Chrome OS or the browser?
mgutz 22:01 chromium browser
taxilian 22:01 oh, good =] that should work fine, then. you're already familiar with the difference between a plugin and an extension? (sounds like you do, but it's the most common misconception in the business)
mgutz 22:01 basically, we want HTML5 + CSS3 presentation but we need to hook into OS
we can't do what we want with the Pepper API
taxilian 22:01 heh. pepper is dead; it's successor won't do what you want =]
kinda too bad; pepper had some potential
you probably don't want a UI, then, right?
at least not one drawn by your plugin
mgutz 22:01 i haven't done C++ for a long, long time .. this should be fun. I hope it's like riding a bike
taxilian 22:01 hehe. I can recommend some good books, if you're interested; mainly, though, the #1 thing you want to make sure you understand to use FireBreath is how boost::shared_ptr is used
as a convention, everywhere in FireBreath where you find a ObjectNamePtr it's a typedef for boost::shared_ptr<ObjectName>
mgutz 22:01 k
taxilian 22:01 and it's really key to not getting yourself hosed. if you do it right you don't have to directly worry about any memory management; just use smart pointers to handle it for you
mgutz 22:01 All UI will be in HTML, Javascript
Plugin will be for sockets, file system access
i hope that's possible
taxilian 22:01 yep. are you only targetting Chrome, or did you want to support other browsers as well?
mgutz 22:01 no, it's meant to be a desktop app
we were goign to use AIR, but decided Chrome makes more sense
taxilian 22:01 okay, shouldn't be a problem, then. IE protected mode can be a pain, but Chrome should be good
mgutz 22:01 awesome :)
FB_GitHubBot 22:01 FireBreath: firebreath-1.4 Richard Bateman * 577dfce (1 files in 1 dirs): Added FB_GUI_DISABLED tip to fbgen tpl -
FireBreath: master Richard Bateman * 577dfce (1 files in 1 dirs): Added FB_GUI_DISABLED tip to fbgen tpl -
taxilian 22:01 to make sure you don't have to deal with drawing stuff, uncomment the set(FB_GUI_DISABLED 1) line from the pluginconfig.cmake generated by fbgen. If you've already generated your plugin, copy it from fbgen/src/PluginConfig.cmake
I just put it there =]
that'll disable drawing and should make your plugin windowless
mgutz 22:01 cool, i was goign to try it on Mac in a bit
taxilian 22:01 it's still a little experimental at the moment, but it will be solid by the release
with Mac make sure you disable all the drawing models
mgutz 22:01 when is released slated?
taxilian 22:01 well, the definite answer is "when it's ready"
mgutz 22:01 lol
taxilian 22:01 my guess is that we'll have an RC by the end of the month
and a full release by mid-Feb
but I'll push it up if everything is ready before then and I'll delay it if we need to
I'd rather have an extra beta or release candidate than release something that isn't solid
actually the beta is pretty good right now
it's mainly the newer features that still are finding bugs and tweaks
mgutz 22:01 is firebreath used at your company?
taxilian 22:01 I was hired for my current job specifically to work on FireBreath
and I was contracted to write it originally
I have in the mean time worked both at places where we used it and places where we didn't
mgutz 22:01 nice
taxilian 22:01 it's been an interesting ride
sometimes I feel like nobody else is contributing, then sometimes things just fly because people are finding and/or fixing things all over and I'm just directing traffic
it's a lot of fun
mgutz 22:01 I researched the past few days on how to implement our plugin. Did not see much on firebreath
I don't remember how I stumbled upon it
taxilian 22:01 where were you reading?
mgutz 22:01 Most of my searches started with chromium npapi
that led me to pepper
taxilian 22:01 amazing how much information there isn't about plugins, no/
mgutz 22:01 true
taxilian 22:01 yeah; we've tried to put up links where possible:
but we can't force people to link to us
I really would like to see the chromium npapi page put a link to our project
and to the couple of others that exist for that matter
mgutz 22:01 they dissuade use of npapi in favor of native client
taxilian 22:01 right
because it's safer
mgutz 22:01 yep
taxilian 22:01 which is exactly why it doesn't work for what you need =]
mgutz 22:01 yep
i wish chromium had a project like Firefox Chromeless
taxilian 22:01 huh. hadn't seen that before
mgutz 22:01 i have to figure out how to get rid of the address bar and tabs
taxilian 22:01 is one of the places a lot of people find out about FireBreath
mgutz 22:01 unfortunately, gecko, in our tests is noticeably slower for a heavy javascript app
taxilian 22:01 this is true
thought I do wonder how ff4 will measure up
well, I'm headed to bed. almost 11pm here =]
good luck