IRC Log Viewer » #firebreath » 2010-10-29

IRC Nick Time (GMT-7) Message
neilg_ 07:10 Morning/afternoon all!
cygmatic 08:10 hi neilg_
taxilian 09:10 anyone think of any reason not to release 1.3?
kylehuff 09:10 I am an ardent obstructionist?
thats all I got...
taxilian 09:10 good enough
we'll put it off a month
oh, wait, that was kyle
now, if *neilg_* were an ardent obstructionist, then we'd put it off
kylehuff 09:10 well, thats put me in a bind too, because what I if I want to obstruct obstruction? that is hardcore..
taxilian 09:10 ouch
let me know what you decide
kylehuff 09:10 I'll sit this one out, I have little motivation to do anything ATM...
neilg_ 09:10 I'm the leader of the Ardent Obstructionist Party. Elect me next Tuesday! lol
taxilian 09:10 lol
neilg_ 09:10 Also: The rent is too damned high
kylehuff 09:10 lol
taxilian 09:10 going to try to fix the "spaces in path" bug with prep scripts, then I'll release
neilg_ 09:10 I think that's fine, nobody else but me has had issues (or really used?) BrowserStream objects
Those are the only real issues I know about
taxilian 09:10 yeah; we'll definitely try to get your fixes in 1.4
but 1.3 already has a *lot* of stuff in it
neilg_ 09:10 Right, it makes sense
taxilian 09:10 there is also a possible issue with using system boost on windows, but there is a simple workaround, and few people use that feature
kylehuff 09:10 what is "system boost" ? I am using boost::thread on windows, not related?
taxilian 09:10 system boost is when you use the preinstalled boost libs instead of the ones packaged with firebreath
kylehuff 09:10 ah
I'm just using the one with firebreath
taxilian 09:10 yeah, it's almost always easier
which is why I go through all the pain required to get it in there
personal pet peeve: I hate projects where I have to spend hours setting up my box before I can build it
kylehuff 09:10 yeah, me too, but that rarely happens since I use debian.. 9 out of 10 times it only requires `apt-get install <whatever I possibly need>`
neilg_ 09:10 You know what I hate? IE. Specifically IE Protected Mode. It made me go from very happy to very sad today. :(
taxilian 09:10 lol
kylehuff 09:10 which is why it is such a godsend to be able to cross-compile.... someday I will be able to do that with firebreath I hope...
taxilian 09:10 need help solving a related problme? I have more experience with IE protected mode than I would like.. (which technically is any)
kylehuff 09:10 neilg_: I'm with you there.. IE can die slowly..
taxilian 09:10 if you guys are any good with activex, we could work towards it one peice at a time...
neilg_ 09:10 It breaks our plugin in IE as it makes everything our plugin loads run with low-user rights
taxilian 09:10 yes, yes it does
one possible solution
that many use
kylehuff 09:10 I am sad to say in this case (because it is not much help), that I am proud to say, I know nothing about activex
taxilian 09:10 is to launch an external process
neilg_ 09:10 It also virtualizes where everything loads from so it doesn't work so well
taxilian 09:10 kylehuff: I too was once happy and innocent
then I had to learn activex to write a plugin
so I vowed that some day I would write a system to allow people not to know activex
neilg_ 09:10 Right, I was looking at that as a possible solution. We're also looking into whether signing everything makes any difference
kylehuff 09:10 some things you can't un-see.. activex, goatse... the list is long unfortunately..
taxilian 09:10 sure doesn't, but signing is a good idea
neilg_ 09:10 What's really upsetting is if we load our DLLs from inside Program Files then everything works
taxilian 09:10 ?? that doesn't make any sense
neilg_ 09:10 Right. It's messed up!
taxilian 09:10 I guess it probably just doesnt' virtualize program files
neilg_ 09:10 But it's entirely true - and that's what I guess too
taxilian 09:10 but it should still keep you from writing anything
if it is helpful, you can set up a named pipe connection between a low and medium integrity process
neilg_ 09:10 And that's fine because we don't write anything. But we can't update content (and DLLs are part of the content...) if it's in Program Files without admin rights
taxilian 09:10 right
neilg_ 09:10 I think Window handles are global so that may well be what we have to do
taxilian 09:10 so to update the content, you can download using hte plugin and then tell your coprocess to move it
it's all very messy
but doable
neilg_ 09:10 That's true, that might be the easiest way. It sucks though. I realise what IE is trying to do but it seems to do it in a very half-assed way
If I can just execute files from within Program Files without issue... It just seems... 'interestingly' designed
taxilian 09:10 totally agree
and that's far from the weirdest issue I've seen
I had one that took Microsoft 3 weeks to figure out
turned out to be an issue with partial keys in HKLM
and was related to our installer; that was the point that I decided that .exe installers should never be used
neilg_ 09:10 I used to work for MS. Great place to work for, lots of amazingly smart people all around the place... But their big problem is supporting legacy code. That's half the reason their security models tend to be broken in significant ways
taxilian 09:10 I guess technically COM in Windows Vista could be considered legacy...
neilg_ 09:10 That's my opinion of course (since I never worked on those systems) but it explains a lot in my mind :)
taxilian 09:10 could be
taxilian 10:10 things like this make me despair for the industry:
iaincollin 10:10 The stupid UK government were pressed to support advocating ditching IE6 (as others, like the German government have apparently done)
but the luddites refused on the basis there was no economic benifit for doing so
(everybody here thinks our government are useless at IT projects, t's an institutional problem regardless of party, we just suck at it)
e.g. we spent 12 Billion GBP on IT project for the NHS, and it was a disaster
for that kind of money the equipment should be capable of taking you to the moon
(sorry that's veering off topic there)
Lots of IE6 here (typically corporate environment)
it's a real hurdle for product demos when you want to show off something, especially at exec/board level (everything works, but's usually not as slick, because it's not worth the effort outside of internal demos)
taxilian 10:10 yeah
iaincollin 10:10 most of the execs and senior staff have Macs here at least, so it's mitigated a bit ...
but then are under pressure to do a Mac port, even though only about 5% of our customers are on Mac :)
taxilian 10:10 hehe
iaincollin 10:10 which is great, I love doing it, but the economics of it are a bit iffy if we don't have a finished Windows build yet :)
taxilian 10:10 yeah, I have definitely been in on discussions like this at companies I have workd at
iaincollin 10:10 Oh I love this quote from the /. comments:
"There's never enough money to do it right, but there's always enough money to do it again."
taxilian 10:10 iain, this spaces in directory thing is a pain. did not expect it to be this annoying...
iaincollin 10:10 oh? I would suggest it's probably not a huge problem for people, maybe could just warn them about it?
taxilian 10:10 I might have to
I hate having things like that in there
iaincollin 10:10 TBH I think that's fine
taxilian 10:10 it just screams "I'm too lazy to fix things, so I'll just make you do things my way"
I have a rather strange set of coding ethics =]
iaincollin 10:10 yeah, usually a bad sign in a project, but overal the rest of it is easily good enough to win people over I think :)
I am looking forward to figuring out why Google Chrome (and only Chrome) is mangling my strings next week
cygmatic 10:10 hm, i have either found a bug in apples webkit plugin wrapper or a major hole in their documentation :/
the sad thing is, the javascript that leads to the problem is too contrived for me to make a clean sample
taxilian 11:10 neither would surprise me
iaincollin 11:10 indeed :/
The FB instability issue with Safari on Mac (but not Firefox or Chrome on Mac...) is a bit suspicous too *glares at Safari icon*
taxilian 11:10 yeah, I keep forgetting to look into that
is there an issue filed?
cygmatic 11:10 whoever has to do cocoa webkit plugins: don't listen to legacy reasons, implement a clean ready function call for the event listener registration or prepare to have your event listeners die on you unpredictably
taxilian 11:10 cygmatic: are you doing an actual webkit plugin as opposed to an npapi plugin for webkit?
cygmatic 11:10 one of the plugins i maintain is cocoa/webkit
taxilian 11:10 I would love to find a way to add webkit plugin support to FireBreath
cygmatic 11:10 what for?
taxilian 11:10 it would perform better when running on Safari
cygmatic 11:10 i don't think it allows for anything that npapi doesn't cover
taxilian 11:10 gives you better access to graphics APIs
cygmatic 11:10 hm, but it's still a third arch in the code-base
taxilian 11:10 true
but it would be awesome :-P
iaincollin 11:10 hehe
cygmatic 11:10 hm, the biggest problem would be that the "supporting all from one binary/bundle" would break
webkit plugins have bundle type WBPL
taxilian 11:10 this is true; we'd have to create a seperate project type for that
cygmatic 11:10 apart from that it shouldn't be too bad if someone HasTheTime(tm)
taxilian 11:10 hehe =]
I'll put that on my list
right after removing the ATL dependency
and making cross compiling work
cygmatic 11:10 removing the atl dependency would also get us closer to getting FB more library like
taxilian 11:10 very true
cygmatic 11:10 the atl rgs registration is one of the main blockers i see for that
taxilian 11:10 right
I'm considering a JSON file format to replace it
cygmatic 11:10 one way i guess, the format doesn't really matter once the registration code can do the same as ATL :)
taxilian 11:10 right
I'm just not sure I want to write another parser. I mean, it's not really that bad using the boost tokenizer and all, but still
I just get tired of maintaining things like that
cygmatic 11:10 i know, json or whatever JustWorks(tm)
taxilian 11:10 right
unfortunately, XML would probably be the more appropriate format
but I don't like XML... :-P
cygmatic 11:10 actually i don't like it either ^^
especially with all those people saying that it obsoletes all other formats
taxilian 11:10 right
cygmatic 11:10 it's just too heavyweight and/or too cumbersome for many use-cases
taxilian 11:10 for most things (IMO) it is just unneccesarily verbose
in this case, though, it would work well
now there is an idea
we could use basically the same format that WiX does
ok, the project generates when spaces are in the path now...
on windows
as soon as I put the change in trunk
now to see if it builds
neilg_ 11:10 Hey, does anybody know anything about the information bar in IE? Specifically is there something I can do to stop that appearing the first time somebody goes to a website that uses a plugin for the first time?
taxilian 11:10 neilg_: did we not have this conversation already? maybe that was wtih iaincollin...
neilg_ 11:10 Nope, not me!
taxilian 11:10 I really need to write this down
there are a few things you can do that help
but I don't know a way to catch everything
iaincollins 11:10 yes I asked about that too :)
I have a solution yes, let see
taxilian 11:10 excelent... I don't have to go track it down again :-P
iaincollins 11:10 You can add it per site, or globally
You only need it for MSIE >6 (IE 6 doesn't care)
FireFox also displays an annoying yellow bar, but in it's case if it can't find a plugin for the object's type (I am using JS dynamic object creation to avoid this)
that's just as we manage the site the plugin will be on in our case, I would guess integrating Firefox's plugin finder support would be the way to go in most cases
it would probably be useful to have that in the default WiX installer (am going to suggest it once I've done it and had it tested, possibly with a couple of other minor changes just for an even better experience)
neilg_ 11:10 Awesome, thanks
iaincollins 11:10 I do love that Microsoft keep doing "here is this annoying this we've done in the name of security; and here is how to get round it" articles
taxilian 11:10 yeah, it's great
iaincollins 11:10 (e.g. UAC, ClickOnce, ActiveX...)
unprivilaged apps talking to privilaged services is a pretty common one (that way you only need UAC at install them, then you can secretly go nuts anytime thereafter)
taxilian 11:10 cygmatic: do you have a minute to look at issue #88 on mac?
cygmatic 11:10 will be back in an hour
taxilian 11:10 you know what, guys? windows batch files are *really weird*
the fix for issue #88 on windows was to add extra " in several places
needed ""path"" instead of just "path"
still have no idea why
neilg_ 11:10 In a batch file "" = "
Because normally a single " is discarded and is a hint to the shell that the contents of the string are one "word"
I have no idea what the bug is or whether that makes any sense to you but...
I'm with you that batch files are weird though ;)
taxilian 11:10 it sorta does
but there are times when "" doesn't work and it has to be "
neilg_ 11:10 It depends on whether what you're calling expects the quote or not. But I agree, it's not consistent
It all comes from copying the old CP/M syntax and not really improving on it too much...
taxilian 11:10 huh
well, now I know who to ask next time I'm trying to do batch file stuff ;-)
I'd really like to overhaul the prep script stuff; make something a little cleaner
iaincollins 11:10 aah yes I've found that "" too
has driven me crazy
neilg_ 11:10 I've been coding batch files since '93 at least! Which is just craziness, where does time go?!
iaincollins 11:10 I still can't work out how shell escaping with ^ in Windows is supposed to work (but quoting stuff seems to get round it)
neilg_ 11:10 And then I used to use CP/M before that on my old Amstrad CPC6128
I've mostly switched to Python these days :)
taxilian 11:10 neilg_: believe it or not, my first batch files were probably before that, but since I was 7 at the time I didn't do anything incredibly advanced, and I've had a few years to forget it...
I used to make menus with choice and if errorlevel statements on my 8088
iaincollins 11:10 those Amstrad's were suprisingly awesome, used to make games in basic, pretty decent graphics support on them
(relatively speaking)
(compared to how sluggish they were at the console anyway :)
taxilian 11:10 =]
iaincollins: issue #88 is fixed on windows, I think it's fixed on mac as well; just doing a bit more tsting to be sure
and as soon as I have confirmed it is fixed and push the fix up, I will release 1.3
... and it's building
the release
cygmatic 12:10 taxilian, you were saying?
taxilian 12:10 cygmatic: no worries, I got it
1.3.0 is released
I'm just updating the downloads (deprecating old ones), then I need to add one thing to the version history
and then I'll send out the notice
putting out a release is kinda a lot of work :-P
cygmatic 12:10 ok, great
oh, i forgot to check the 10.5 issue at work :|
next week then
taxilian 12:10 yeah, I saw that
but it's something with an easy workaround
so not worth holding the release for
if there are enough issues we can do a 1.3.1
cygmatic 12:10 i have no idea about it, i'm only guessing it's due to 10.5 forcing you on older xcode versions which don't have that entries that are patched
taxilian 12:10 lol.. and someone has already downloaded it. was that one of you guys? has the release file
well, that seems possible, except that I was looking at xcode 3.1.x project files when I wrote the patch tool
and I know we've had it working with xcode 3.1.x
cygmatic 12:10 right, cmake 2.6 issue with 3.1.x files then maybe?
taxilian 12:10 most likely
but what specifically I don't know
changelog is updated
cygmatic 12:10 looks good
by the way, can we maybe get notifications for wiki changes like with the gcode wiki?
taxilian 12:10 you would think so
but I haven't figured out how to do it yet
cygmatic 12:10 hm, this looks good as a per-user setting:
taxilian 12:10 ahh, yes
I think we'll leave it with that rather than send more email to the scm list
but I do have email sending working (I think) so this should work
cygmatic 12:10 hm, i don't see the "browse space" thing though
ah well, will search later
nirvdrum 12:10 taxilian: Your tweeted link doesn't work.
taxilian 12:10 strange; it dropped the ! at the end
ahh, well
figures. someone finally finds an issue that should have been fixed in 1.3.0... minutes after I do the release
ahh, well. it's a pretty minor leak
be back in a few
iaincollins 14:10 hehe .. congrats all!
I'll hold off for that upgrade then :)
I <3 "no breaking changes"
oh since RC, but the others are minor and nicely documented, so that's great
have a great weekend and thanks again @cygmatic, @taxilian_away, @amackera_away
taxilian 16:10 I was just contacted by a large company that may want to start using FireBreath for their plugin
cygmatic 17:10 again? nice
you know, if citrix writes you remind them that i already work for them ;)
taxilian 17:10 lol. this company is the company I worked for when I first started the project =]
sounds like they'd like to hire me, but unfortunately I'm already employed
but they still want me to come in and give them a presentation of why they should use FireBreath
I guess it's not a large company in terms of Facebook
just in terms of the potential install base of the resulting plugin
cygmatic 18:10 ah. i guess you'll convince them ^^