IRC Log Viewer » #firebreath » 2010-10-23

IRC Nick Time (GMT-7) Message
cygmatic 03:10 taxilian: where does an X11 plugin link to PLUGIN_INTERNAL_DEPS?
ah, projectDef
taxilian: do you have a moment to look at a linux linking issue
cygmatic 05:10 well, if you can: http://code.google.com/r/georgfritzsche-fb-factory-trunk/source/checkout
i'm stuck and don't see it
taxilian 09:10 cygmatic: I'm up now
cygmatic 09:10 great, i'm almost through looking through new plugin questions on SO ^^
taxilian 09:10 lol
=]
let me knwo if there are any I should look at
and don't be shy about shamelessly pointing people to FireBreath as appropriate… ;-) stackoverflow is still one of our best sources of traffic =]
cygmatic 09:10 i know, still doing it - mostly with a little disclaimer :)
taxilian 09:10 well, yeah
that's fair =]
cygmatic 09:10 i've got the factory working on windows and mac now, but on linux i can't get around those 2 linking errors
taxilian 09:10 ok; I'm pulling it down on a linux machien
cygmatic 09:10 cool, i might just be routine-blinded
so a 2nd pair of eyes hopefully helps
taxilian 09:10 hopefully =]
cygmatic 09:10 one guy wrote on SO regarding IE: "Around 2012 Microsoft will stop supporting Active X"
i'm quite curious where he got that from
taxilian 09:10 yeah; that seems somewhat… unlikely...
'course, people keep saying that all the browsers are going to stop supporting plugins entirely
I just don't see that
though they might create a new plugin interface to try to be more secure
cygmatic 09:10 yeah, as long as dynamically loaded content doesn't have full access to the system, there is a need
they will cover more with OGL extensions etc., but not everything
taxilian 09:10 so I'm getting 4 link errors
cygmatic 09:10 yep, those are all implemented though
taxilian 09:10 ok
looking into it
where are they implemented?
cygmatic 09:10 getPluginName(), getPluginDescription() are in PluginCommon/PluginDescription
Npapi::createNpapiPlugin() is in NpapiPlugin/X11/NpapiPluginX11.cpp
createPluginWindowX11() is in PluginWindow/X11/PluginWindowX11.cpp
the same approach works fine on both other platforms, so i guess i'm either overlooking something stupidly simple or the cmake config is messed up
taxilian 09:10 ok; hang on
I don't really have this machine set up for development, so I have to install my tools on it =]
cygmatic 09:10 sure :)
ok, using "nm -C" to check the unresolved symbol from libPluginCore.a vs. the exported one from e.g. libFBTP_NpapiPlugin.a looks good too
taxilian 09:10 hmm. the projects are definitely listed in the link libs
kalev: you around?
kalev 09:10 yep
taxilian 09:10 do you have a minute you could look at a linking problem on linux that is stumping both Georg and I?
we really need this working soon...
(actually, that should be Georg and me, I suppose)
kalev 09:10 sure, where can I get the source?
taxilian 09:10 http://code.google.com/r/georgfritzsche-fb-factory-trunk/source/checkout
kalev 09:10 okay, I'm on it
taxilian 09:10 and I promised my sister I'd help her with some stuff and I have to leave now :-/ if you guys can't figure it out I'll look at it again this afternoon
I'm pretty certain this is something really simple and small
but I'm not seeing it
thanks, kalev
cygmatic: be back later this afternoon :-/
cygmatic 09:10 ok, see you later
kalev, if you have questions, ask away
kalev 09:10 cygmatic: I cloned the treen and now I'm building examples
src/NpapiPluginTest/NpapiPluginTest.cpp:38:1: error: ‘HINSTANCE’ does not name a type
that looks like a win32 API type
cygmatic 09:10 ooh
thats a new one
can you try to build FBTestPlugin in the meantime
i know what i broke for the windows part earlier :(
kalev 09:10 sure.
cygmatic 09:10 fix for windows-leftovers is pushed
cygmatic 09:10 and one for missing link flags for the npapiplugin test
kalev 10:10 cygmatic: looks like the problem is in the order in which the static libraries appear on the linker command line
cygmatic 10:10 ah, that would make sense
kalev 10:10 apparently the order matters and I managed to remove some of the linker errors by just moving PluginCore a little bit up in the PLUGIN_INTERNAL_DEPS list in cmake/projectConfig_main.cmake.in
have you seen http://stackoverflow.com/questions/45135/linker-order-gcc ?
looks like you have a circular dependency between PluginCore and ${PLUGIN_PREFIX}_NpapiPlugin
cygmatic 10:10 yep, there is :/
i didn't think this would make problems
hm, i don't know wether this can be resolved to a well-defined order
thanks for it finding it though :)
kalev 10:10 hope it helps
cygmatic 10:10 definitely, i finally know where to look
taxilian 12:10 cygmatic: Any luck?
cygmatic 12:10 just now it linked :D
at least FBTestPlugin
now trying a "make all"
taxilian 12:10 cygmatic: awesome! good work
cygmatic 14:10 taxilian: should be done, would you confirm with a quick test?
then i could merge it in
taxilian 14:10 I can do a build; not in a position to do any real tests, though :-/
cygmatic 14:10 quick fbtestplugin page opening?
taxilian 14:10 I'll see what I can manage
what is the URL again?
have I mentioned how happy I am about this change?
cygmatic 14:10 http://code.google.com/r/georgfritzsche-fb-factory-trunk/source/checkout
you did now :)
besides the cyclic dependencies it doesn't look too bad either
taxilian 14:10 well, we have so many of those one more won't be a huge deal
cygmatic 14:10 maybe getting rid of the ifdefs would be nice, but that would mean ripping it appart in 4 classes
taxilian 14:10 yeah
I prefer the ifdefs
and you know how much I hate those =]
cygmatic 14:10 heh, point taken
taxilian 15:10 thing is, they can still do platform specific implementation in their code
so that they don't have any #ifdefs in their code
cygmatic 15:10 not for the class definition itself
but yep, could still be split apart or straight forwards to actual implementations in ${Platform}/
taxilian 15:10 right
and they wouldn't have to see the class definition itself
oh, have you added docs for it?
if you haven't, I can probably find time sometime; you've spent a lot of time on this
cygmatic 15:10 i will add some @briefs
but then i'll be in bed ;)
taxilian 15:10 lol. okay =]
building now...
every time I build from scratch I'm tempted to figure out how to use precompiled headers on windows
from what I understand, it would give a huge boost to compile time
cygmatic: works for me so far
on windows
woah… wait. crash on IE
cygmatic 15:10 hm, great
i think i overlooked testing in IE
taxilian 15:10 hmm. I think it's my fault, not yours
yup, sure enough
cygmatic 15:10 oh?
crashes for me too
taxilian 15:10 yeah
but it's a change that I made that crashes it
glad you made me try this… wanna kick myself. I thought I'd tested that change
cygmatic 15:10 happens, especially if the day is long enough
what do we do?
taxilian 15:10 I'm fixing it
then I'll put the fix in stable
and you can merge it in
cygmatic 15:10 ok, then i'll merge it sometimes late tomorrow
taxilian 15:10 tell you what
I'll test this a bit more
and merge it to head
add some docs
and you can polish it up when I'm done
that way there will be a two-way code review by default
cygmatic 15:10 sounds good, i'll look over it tomorrow evening then
taxilian 15:10 sounds great
get some sleep
thanks for your work
this is awesome
cygmatic 15:10 good night then :)
taxilian 15:10 sleep well and wake =]
cygmatic 15:10 thanks, you too
taxilian 16:10 just a heads up to everyone, I just pushed cygmatic's refactor of the factory stuff to live. That will be a breaking change
but it should be pretty easy to update to
taxilian 18:10 I am overhauling the fbgen template; adding a lot more documentation, weak_ptr to the plugin object, etc
taxilian 21:10 building the rc2 archives now...
wow. just barely uploaded it, and someone has already downloaded it