IRC Log Viewer » #firebreath » 2012-09-19

IRC Nick Time (GMT-7) Message
aa 01:09 aa
JuanDaugherty 06:09 prolly my bad, but seems like the plugin gets instantiated twice in IE
JuanDaugherty 08:09 on FF turns out there's only 1 plugin container per plugin so you can you can reload your test page after attaching to it
JuanDaugherty 08:09 but that's really unsatisfactory
taxilian 08:09 that the plugin isn't unloaded between page requests?
JuanDaugherty 08:09 no that i can't get control the first time it loads
taxilian 08:09 I already told you a couple of ways you could do it, and you suggested one yourself
granted, it's not ideal, but it's possible
JuanDaugherty 08:09 the best would be if your stuff supplied the missing FF plugin startup dialog
taxilian 09:09 hehe. well, that's an interesting idea; I'm not sure how we'd detect that you wanted the dialog… maybe a file on the system, or a reg key, I guess. anyway, I welcome pull requests
JuanDaugherty 09:09 so what were you're suggestions again?
taxilian 09:09 the one I usually use is to just add sleep(10000) to NP_Initialize in np_winmain.cpp (in PluginAuto)
but your idea of an assertion would probably work too
JuanDaugherty 09:09 no that didn't
but the sleep will
you know I think for debuggng MS has the best IDE
taxilian 09:09 easiest to understand, anyway
JuanDaugherty 09:09 don't really have enough experience of xcode, but suspect it's similar to eclipse, also uses gdb
taxilian 09:09 xcode used to be really nice
and still is for objective c
but they've totally hashed the c++ support in the last few versions
but it takes longer to get good at using
JuanDaugherty 09:09 the super irritating thing is vcpkgsvr
taxilian 09:09 yeah; that being such a well-known issue, you'd think they'd fix it
JuanDaugherty 09:09 i'd like to kill somebody for that
you're trying to attach and the damn thing is snagging the attach dialog. Also it's not generally sufficient to kill one instance
JuanDaugherty 10:09 taxilian, ping
taxilian 10:09 JuanDaugherty: pong, your latency is about 30 seconds
JuanDaugherty 10:09 so I've gotten the FF snafu cleared, you need also to reset the plugin container timeout btw
taxilian 10:09 ?
JuanDaugherty 10:09 everything looks OK, but I'm wondering
when a mime type is loaded is there something you have to do to cause the rendering to be invoked?
taxilian 10:09 nope
JuanDaugherty 10:09 was you question about the timeout? The plugin container in FF 15 is set to timeout in 45s if the plugin doesn't load
taxilian 10:09 ahh
that's new
JuanDaugherty 10:09 btw I'm just exiting if the destructor gets called before anything has really happened, just skipping the stuff you had there still traps
JuanDaugherty 10:09 how is the developer supposed to work the NP saved data without modifying your code? It looks like you are just ignoring the npsaveddata?
taxilian 10:09 never needed to use it
so never wired it in
pull requests are welcome =]
JuanDaugherty 10:09 you could just abbreviate that PRAW
so this is the forking issue
taxilian 10:09 hehe
well, I think you could actually still override that method
but I'm not 100% sure
I'd have to go look at it
JuanDaugherty 10:09 probably the plugin I'm working is more of a workout of your package than it's gotten so far
taxilian 10:09 you're the first to ask me about npsaveddata
JuanDaugherty 10:09 it's a pretty basic thing
taxilian 11:09 thinking back, I think the main thing was that there was no equiv on IE that I found
so I don't use it for anything
thus it isn't in the framework
my use cases tend to get priority over anyone else who isn't willing to write it themselves :-P
JuanDaugherty 11:09 2bsure
but this plugin runs in IE with same logic
taxilian 11:09 if you can get the info needed to support npsaveddata dn the equiv on IE and then put it in a jira issue I'll probably get to it eventually
or PRAW ;-)
JuanDaugherty 11:09 :)
reichi 11:09 taxilian: do you know what will happen once wayland find's it's way into linux distributions?
taxilian 11:09 wayland?
reichi 11:09 and x11 replacement
taxilian 11:09 not a clue
reichi 11:09 which under heavy construction currently
JuanDaugherty 12:09 just a comment about PCH. Probably the main negative is that it kills intellisense, at least outside of your dev env.
taxilian 12:09 yeah; I'm going to have to experiment with that. it is very strange, because pch should normally drastically improve things like intellisense
JuanDaugherty 12:09 well you don't seem to use it in the normal way
I assume due to x pltfrm/cmake
when you said it improved performance 20% did you mean that it cut it by amount or did you mean it cut it 80%?
s/performance/compile time/
taxilian 12:09 I don't remember the exact numbers, but it decreased the build time to about 20% of the old build time
JuanDaugherty 12:09 thought that's what you meant
taxilian 12:09 off to lunch, bbl
JuanDaugherty 12:09 wait quick q!
when a plugin is created there are some argc type parms
where do they come from?
vinay_ 13:09 Hello, are logs configured for console (through FBLOG_xxx) supposed to show up in the browser console? They aren't showing up for me. Any suggestions?
taxilian 13:09 no
they are not
vinay_ 13:09 Does printf send it to browser console?
taxilian 14:09 no
m_host->htmlLog does
but don't use that in production; there are some significant performance issues with it
vinay_ 14:09 Thanks
JuanDaugherty 14:09 from embed tags apparently.
taxilian 14:09 JuanDaugherty: sorry, I didn't see your question
yes, it comes from the object or embed tag
I find I have fewer random issues with object tags, so I don't use embed
(I don't rightly remember what issue I had, but there was some doozy with events I think that went away when I switched to an object tag)
JuanDaugherty 14:09 so the object tags are exposed as js objs I take it
taxilian 14:09 kinda a weird hybrid, actually
object/embed tags have html attributes, but it queries the NPObject before exposing anything
i.e. if you access .id, it asks the NPObject if it has a property called that first, then if not it asks the DOMElement
it can get a little weird if you aren't careful
JuanDaugherty 14:09 1 weird trick:)
haochen 14:09 Hi does anyone here tried to read registry key from a firebreadth plugin?
taxilian 14:09 sure
haochen 14:09 i have a accesss problem with that
taxilian 14:09 which browser?
haochen 14:09 firefox
taxilian 14:09 okay; the problem is most likely that a normal user process can't acccess that part of the registry, then
haochen 14:09 i checked the key in registry, and it has read permission for normal user
i'm using RegOpenKeyEx(HKEY_LOCAL_MACHINE, TEXT("..."), NULL, KEY_ALL_ACCESS|KEY_WOW64_32KEY, &hKey);
but i got access_denied error
taxilian 14:09 try making a quick standalone app, see if you can still access the key the same
because there is no sandboxing or anything that would affect that that I am aware of
you can also use process monitor to monitor what is happening to the registry and that can be helpful in tracking down issues like that
haochen 14:09 okay thank you
I'll try that then
the process monitor tells me ACCESS_DENIED, while its desired access is "All Access"...