User:Sayrer:OinkNotes

tarball from taras ssh keys my oink bug

taras>	hey crowder
	<taras>	word on the street is that oink runs on osx these days
	<sayrer>	taras: is your stuff available anywhere?
	<sayrer>	been meaning to check it out
	<bent>	anyanyone know how to do decodeURIComponent from C++?
	<Mano>	josh: ?
	* bent	isn't so great at tracing through the js engine
	<taras>	sayrer: ya
	<taras>	http://people.mozilla.com/~tglek/dos_and_squash-0.0.tar.gz
	<taras>	i just updated the tarball
	<gavin>	bent: that's a native method implemented in spidermonkey
	<gavin>	you could just copy it, I guess...
	<gavin>	or you could use the (necko?) unescaping interfaces, I guess
	<gavin>	though you'd not be garanteed the same behavior
	<taras>	sayrer: docs are pretty much in blog posts and --help
	<taras>	let me know if you get confused or lost in all that
	<sayrer>	taras: does the js stuff work yet?
	<taras>	i'll see if i can package it better
	<taras>	define js stuff?
	<taras>	you can do function-local analysis with it
	<taras>	for C functions, i haven't tested C++
	<taras>	but feel free to do so
	<sayrer>	for some reason I thought it was going to be possible to write JS handlers to translate specific parts of C++ to some other syntax
	<sayrer>	is that right?
	<taras>	no that's a big todo :)
	<taras>	tell me what you want to do
	<taras>	perhaps i'll do that next
	<taras>	i have been trying to figure out how do js guided c++ transformations
	<sayrer>	taras: I wanted to use your thing to try simple conversions from C++ to JS. maybe just .h files to js stubs.
	<taras>	sorry dude
	<taras>	that was on the wiki
	<taras>	for possible applications of oink
	<sayrer>	not going to happen?
	<taras>	it is going to happen
	<sayrer>	oh, "not yet" is ok with me
	<taras>	brendan really wants it
	<bent>	gavin: this? http://lxr.mozilla.org/seamonkey/source/xpcom/io/nsEscape.h
	<taras>	can you give me an example of the .h file you want to translate?
	<taras>	i might be able to whip up a quick tool if it's easy
	<taras>	been looking for an excuse to get back into olmar
	<sayrer>	taras: anything generated from these:
	<sayrer>	seamonkey
	<sayrer>	er
	<sayrer>	http://lxr.mozilla.org/seamonkey/source/editor/idl/
	<taras>	so you want to generate js from idl?
	<sayrer>	as a first step. if I can venture into generating implementation too, I will
	<sayrer>	but I might want to generate from the .h files that include these
	<sayrer>	the implementation classes
	<sayrer>	so I get all their private methods
	<taras>	ah
	<taras>	that's relatively easy
	<taras>	i mean, generating stuff from headers
	<sayrer>	is it a good way to learn your tool?
	<taras>	my tools are overkill for this task
	<taras>	i can just help you write your own
	<taras>	it's relatively easy
	<taras>	just have to get used to oink which my tools are based on
	<sayrer>	ok, that sounds good.
	<sayrer>	since I am oink-clueless at this point
	<taras>	ok
	<taras>	so you are interested in doing that?
	<sayrer>	sure
	<sayrer>	I will file a bug
	<taras>	ok
	<sayrer>	what is your bug mail
	<taras>	tglek@mozilla.com
	<taras>	you should get an account on the oink machine then
	<taras>	for convenience
	<gavin>	bent: I guess so, yeah. I'm not very familiar with any of that code.
	<gavin>	biesi would be the guy to ask, I think
	<bent>	gavin: will do, thnks
	<RyanVM>	what does "* > form" mean in a css sheet?
	<ajschult>	RyanVM: a FORM tag that's a child of another tag
	<RyanVM>	thanks
	<Jesse>	as opposed to a form that's the root element?
	<ajschult>	Jesse: apparently :)
	<RyanVM>	hey, I didn't write the stylesheet, I'm just debugging it
	<Jesse>	hehe ok
	<ajschult>	Jesse: or a bastard form :)
	<RyanVM>	check this out: http://www.ryanvm.net/testcases/select.xhtml
	<RyanVM>	try clicking the down arrow
	<Mano>	awesome, five minutes to fire the xcode debugger just for a profile-locked message box
	<Mano>	oh, and a ping-timeout i see.
	<mfinkle>	sayrer: i think you can use xpidl to generate js stubs from IDL
	<sayrer>	mfinkle: i did not know that
	<sayrer>	but I want to venture into oink anyway
	<mfinkle>	well, i might be wrong - it may only be c++
	<bent>	mfinkle: nor did I!
	<sayrer>	c++ it definitely does
	<KaiRo>	rob_strong: thanks for looking at the seamonkey installer patch - are you ok with mxsmurf (Frank Wein) directing further reviews of that code to you?
	<KaiRo>	s/mxsmurf/mcsmurf
	<rob_strong>	KaiRo: I don't mind at all
	<KaiRo>	rob_strong: thanks. we don't have anyone else in our team who knows NSIS, so it's good to have a second pair of eyes on that code
	=-=	Mode #developers +o dbaron by ChanServ
	<rob_strong>	KaiRo: no problem... I figured that would be the case and most of the changes after the initial landing should be fairly simple
	<KaiRo>	rob_strong: yes, I guess so... your other recent field of expertise (Vista support) is probably even harder for us
	<rob_strong>	KaiRo: I'll help out with that as well... I suspect the same headaches will have to be handled with SM too
	<taras>	sayrer: it would be pretty cool to generate JS from C++
	<taras>	could probably reuse parts of dos so you could parse C++ with JS and generate JS with it
	<sayrer>	yeah, I also want to try to generate D from C++
	<sayrer>	but don't tell anyone that
	<taras>	oh not D
	<KaiRo>	rob_strong: yes, we probably have the same problems... I know of default app stuff not working (well), one major problem though is that we don't have real developers who converted to Vista yet
	<sayrer>	js first
	<taras>	i think the best we could do with C++ -> JS would be to generate js dropping all of the pointer, references, etc
	<taras>	so it'd be like lossy pseudo-code
	<taras>	then refine that with pattern matching to optimize for particular api calls or something
	<sayrer>	yeah, could generate unit tests maybe
	<taras>	that would be a good one
	<taras>	did you file the bug?
	<sayrer>	not yet
	<sayrer>	brb

	<taras>	first thing that'd be useful would be to see if oink and friends run on osx
	<taras>	they are supposed to
	<sayrer>	I can try that
	<taras>	sayrer: btw, where are you located?
	<sayrer>	nyc
	<taras>	ok
	<taras>	sayrer: https://bugzilla.mozilla.org/show_bug.cgi?id=356904
	<taras>	this has info about the oink machine
	<taras>	that's where i keep my mercurial repositories
	<sayrer>	ah
	<firebot>	Camino: 'MacOSX Darwin 7.9.0 boxset Depend camino' has changed state from Burning to Success.
	<taras>	i dunno if you need to reopen the bug
	<taras>	as it's set as resolved
	<sayrer>	I just attached my key
	<taras>	as far as oink dev goes
	<taras>	you can start by modifying static print
	<taras>	staticprint
	<sayrer>	ok
	<taras>	override the virtual bool visitTypeSpecifier(TypeSpecifier *obj); method
	<taras>	that gets triggered upon hitting a class
	<taras>	or struct, etc
	<taras>	see squash.cc on how to identify whether it's the class you are interested in
	<sayrer>	I don't see any info on your hg stuff
	<sayrer>	if I log in, will it be obvious?
	<taras>	no, it's not documented
	<taras>	i was just keeping that for myself
	<taras>	that tarball should have a .hg directory so you get all my history, etc
	<taras>	i'll figure out a good place to put the rest of the stuff tommorrow
	<taras>	the machine is useful for actually running squash, etc when lots of horsepower is required
	<sayrer>	taras: ok, no hurry. I have other things to do today
	<sayrer>	I have an opteron here
	<sayrer>	is that enough? or is that thing a monster?
	<taras>	i figure we might as well centralize the oink dev on it
	<sayrer>	ok
	<taras>	for your task anything would do
	<taras>	i'm done for the night
	<taras>	ping me once you get stuck