IRC Log Viewer » #firebreath » 2010-09-23

IRC Nick Time (GMT-7) Message
kalev 07:09 taxilian: hey
taxilian 07:09 yeah, I'm looking into it
I think I know how it happened :-/
should be an easy fix
kalev 07:09 you are reading my mind :)
taxilian 07:09 but I still feel stupid
no, I read your email =]
kalev 07:09 did you use hg export or something?
taxilian 07:09 yeah
fixed it in nightly but forgot to on the release build
kalev 07:09 by the way, I think it's better to not completely remove old tarballs / zip files from the Download section
taxilian 07:09 here is how I do it:
I delete the ones that have never been downloaded
and I deprecate the rest
kalev 07:09 people might want to try older versions and it's usually much easier with tarball releases than with git
I see
taxilian 07:09 I always leave at least one archive from each version
so if you do a search for "deprecated" you'll find them still there
kalev 07:09 oh, I think I was stupid myself
yeah, they are all there, sorry.
taxilian 07:09 no prob
ok, the new archives should be there
I'm sure glad you thought to check that
I probably would have assumed it was a problem with an old version of hg or something
kalev 07:09 glad to help
taxilian 07:09 every day that goes by that our activity doesn't go back down I'm a little suprised
it looks like our push for 1.2.0 did a lot of good
we're averaging over 150 hits / day
kalev 07:09 nice!
taxilian 07:09 that's up 50%-100% from before the push
even Sunday we had over 100
and usually Sundays we get 30-50
kalev 07:09 latest Ohloh hg update appears to have failed:
I hope it's not because of the external boost repo but just some glitch with googlecode
taxilian 07:09 it already worked once with the boost repo
so I changed it back to the stable branch
now that we've pushed the boost change to that
kalev 07:09 ah
taxilian 07:09 it'll take a few minutes (maybe an hour) to run, but it looks like it's working this time
do you want access to update the ohloh stuff?
nirvdrum 07:09 taxilian: Tsk tsk. Should have been 1.2.2 ;-)
kalev 07:09 not really, I was just looking around
taxilian 07:09 nirvdrum: there were no changes in the code
kalev 07:09 and gave you a kudo!
taxilian 07:09 just in the binary
kalev: thanks for the kudo =]
nirvdrum 07:09 taxilian: Yeah, but the release artifacts are different.
taxilian 07:09 the release artifacts are now correct =] you may be right
but I didn't want the changelog for 1.2.2 to say "fixed the release process"
nirvdrum 07:09 Then you shouldn't have screwed up the first one :-P
It's not a big project, so not likely to be a big deal.
At apache for instance, we have to do that.
taxilian 07:09 hehe
nirvdrum 07:09 Just to avoid confusion with people having downloaded 2 different versions of the same binary.
taxilian 07:09 yeah, you're absolutely right
I feel really stupid about that
kalev 07:09 could have worked too
nirvdrum 07:09 Trust me, we all do it.
taxilian 07:09 another thing that I need to figure out
is how best to indicate the version number in the code itself
and how to update that during a build
and for that matter how to update it in source control
at least it's the kind of mistake that should only happen once, since the deploy process is automated
smartest thing I ever did
nirvdrum 07:09 Heh.
kalev 07:09 so, we talked once about removing boost completely from Firebreath mercurial history
taxilian 07:09 yeah
kalev 07:09 I'm not sure if it's a good thing to do as it forces everybody to checkout everything again
but it's pretty easy to do.
taxilian 07:09 that's my worry
kalev 07:09 [[email protected] fb-rewrite-history]$ du -sh firebreath/.hg
31M firebreath/.hg
[[email protected] fb-rewrite-history]$ du -sh firebreath-rewritten/.hg
19M firebreath-rewritten/.hg
so, went down from 31MB to 19MB
taxilian 07:09 hmm. that's definitely nothing to scoff at, though
amackera: we're discussing removing the boost file history
what does everyone else think? may main concern is that it would be potentially dangerous to the integrity of the repo
kalev 07:09 yes, that's mine too
taxilian 07:09 I think we should probably leave it for now
but it is good to know that it is possible
kalev 07:09 I think I would go for this if you ever want to convert the repo to git, but not just like that
taxilian 07:09 yeah
good point
and I'm still considering that
but I've decided there isn't yet a good enough reason to justify it
kalev 07:09 you said that branches in hg are hard to use
are they easier in git?
taxilian 07:09 yeah, I think so
kalev 07:09 I think there's soon going to be a merge hell with dev repo and stable repo
taxilian 07:09 the main thing that makes them confusing in hg IMHO is that which branch you get when you do the clone depends on what was last committed to
lol. well, let's just say that I know about what it's going to require to do that merge
and I'm not really looking forward to it
but I'm not too worried either
it isn't as bad as it looks
kalev 08:09 the problem is that with current setup you will need to merge dev into stable
taxilian 08:09 I can merge stable into dev first
and then push back to stable when we're ready
also, I can merge locally
and not push until I know I have it right
kalev 08:09 with most projects I've worked with people usually create release branches, which are offsprings and will never get merged back into trunk
so all to work goes into linear trunk history
and if they need to release bug fix releases they backport specific fixes back to the release branch
taxilian 08:09 yeah; I've considered that. I would do it to some extent wtih git since I could use git-flow
amackera 08:09 that seems like a very SVNish approach
taxilian 08:09 but I'm going to wait until it is an actual problem before I spend the time required to prevent it from happening again
kalev 08:09 with git it's usually: identify a commit on master, git checkout 1.2-branch, git cherry-pick <commit-id>
cherry-pick takes one commit and backports it, so history in the 1.2-branch diverges from master but flows linearly in the 1.2-branch
and when 1.2-branch has gotten enough fixes, just tag one specific commit and release it as 1.2.2 or whatever
this way there's never need to merge 1.2-branch with master
taxilian 08:09 yeah; I may still switch things up to use that method sometime
but for now, this works
and hasn't been too problematic
I can redo our release system or write new features… :-P
kalev 08:09 yeah, the latter is certainly more important :)
taxilian 08:09 Jarom is looking into improving the FBTestPlugin html test page to add some automatedish tests and then possibly do some windmill integration to make it completely automated
that will be awesome too
anyway, gotta run to school
Anson, if you haven't already, look at the latest issue (72?) with 10.4 support and tell me what you think
kalev 08:09 sounds great
have fun
amackera 08:09 sure i'll check it out
amackera 08:09 go seems like an interesting language
cygmatic 10:09 lol... the ohloh stats on our frontpage say "Total Lines of Code: -253,451"
kalev 10:09 haha
taxilian_away changed it today to index main repo instead of the dev. repo
taxilian 10:09 huh. that's weird
cygmatic 10:09 posted the bug in the forum
taxilian 10:09 I think I'll just leave it as is and hope they fix it
if they don't within a few days, I'll try changing it back or something
it wasn't updating, and I figured it probably should be on the main repo; I only switched it before to be sure that moving boost out was going to have the desired effect
I changed the "wiki" link
let me know what you think
*looks at watch*. I have 10 more minutes and then I'm done with doing FB stuff today -=]
kalev 10:09 Ohloh Analysis Summary: Mostly written in CMake
taxilian 10:09 lol
cygmatic 10:09 kalev, you have been quite busy eh? ;)
kalev 10:09 I'm a busy guy!
cygmatic 10:09 taxilian, can we still access the index somehow?
taxilian 10:09 yes
bottom link on that page
cygmatic 10:09 i like the TOC though
taxilian 10:09 I knew you'd ask ;-)
if you remove the ?m=6 or whatever at the top the edit link reappears as well
cygmatic 10:09 hm, i'll move that up so that i'll stay the only one asking it? ^^
taxilian 10:09 as we get more wiki pages (which is good) it was getting to be hard to find anything
feel free
I just threw it together
I'm sure it could use some tweaks
cygmatic 10:09 and you my personal dynamic list of project suggestions? ^^
+didn't like
taxilian 10:09 actually, I did like it
but I decided it was too confusing
when I tried to add my own entry to it
messed up and created another topic
and couldn't delete it
but if you're really attached to it you can put it back, if you want =]
cygmatic 10:09 no, just asking - i'll move the OGL one to the ideas page
taxilian 10:09 cool
taxilian 11:09
amackera 11:09 sweet :D
nitrogenycs 12:09 I was here
taxilian 13:09 amackera: if you wanted to take a few minutes and write up some more docs sometime, a brief overview of how to use the WiX stuff would be awesome
amackera 13:09 Okay, good call
i'll add it to the TODO
taxilian 13:09 you're the only one who has used it in the last 8 months, I think
amackera 13:09 there's not much documentation available online, i've found
happy to share what i've learned
taxilian 13:09 more than anything else, the fact that it is there should be documented =]
then anything else you can add to make things easier =]
amackera 13:09 wow
i just wasted 1/2 a day because i wasn't aware of the +universal variant in macports
lesson learned: read documentation
amackera 14:09 Is there a way to save changes to a Wiki page without making it public? I can't finish this page yet, but I want to save my progres...
cygmatic 14:09 besides store it as text somewhere, no
amackera 14:09 k
Well i suppose as long as it's not added the main wiki page thing we have now then users won't see it
amackera 15:09 writing documentation is a great way to expose how little one knows about what one is documenting
taxilian 15:09 amackera: yeah, I have to agree with you there
amackera 16:09 well i'm off for now, see you guys
taxilian 19:09 anyone have any spare cycles for a simplish (no c++ required) task?
back in a bit. think about it :-P
cygmatic 19:09 "depends" ;)
taxilian 19:09 hehe
I'd almost rather have you work on code :-P
but on the other hand
this is fairly important, I think
we need to make a better test page for FBTestPlugin
get rid of the alerts
cygmatic 19:09 ok, i'm not really a web-guy
taxilian 19:09 and have buttons and/or automated things that excercise all of the interfaces on the API and test to make sure that they work correctly
really the two best people to do it seem to be Jarom and I
or possibly Ben
if I could talk him into it
though I believe nirvdrum has a lot of automation experience, so he might be a good candidate
the problem is, those people are all as busy as I am =]
that said, I think anyone could improve it from what it is now
cygmatic 19:09 i'd still love to implement the automated testing of plugins in automated browser at some point...
... but i don't see this happening in my spare time
taxilian 19:09 yeah
in the mean time, I think Jarom and I have some good ideas for how we could do it with Windmill
which I think would be possibly even better in the long run
cygmatic 19:09 what is that?
taxilian 19:09 though the automated browser could have more uses when you start tweaking it
simply because then we could test in the actual real browser on each platform
and potentially it could be done in a reusable way
to help others do test harnesses for their plugin
cygmatic 19:09 yep, the real browser was the point
taxilian 19:09 maybe we're talking about the same thing, then
I was thinking of an idea you and I discussed once to basically emulate a browser to test all of the stuff
cygmatic 19:09 i was thinking of using embedded controls (webkit, iwebbrowser2-stuff, ...)
i.e. the real browser-engines
but you are talking of automating the actual end-user apps?
taxilian 19:09 yeah
on a build machine
rather, a VM
cygmatic 19:09 hm, how would you set up test-cases then?
taxilian 19:09 basically things like "Add the plugin to the browser. wait 5 seconds. Did the event fire?"
"pass this array into this function, look at the output
cygmatic 19:09 ok, i don't know how such testing is done today
taxilian 19:09 I don't really either
but Jarom does
and he's probably the best QA engineer I know
and he says it's doable
cygmatic 19:09 if that works easily its definitely nice
taxilian 19:09 the hardest part is finding the time
and I think the first step is having a page that can be loaded that will basically run its own set of tests on FBTestPlugin
and you just look at it to see if it worked
because right now we don't have an easy way to test all features
but, one thing at a time
cygmatic 20:09 hm, i just realized that i wouldn't even need to transform the testing javascript for my testing approach
though it would be much nicer to use with transformation
taxilian 20:09 so is there any chance your company will switch to FireBreath now that it has a friendly license and good mac support?
cygmatic 20:09 no, after our take-over we are now focussing on the core engines only
taxilian 20:09 ahh
cygmatic 20:09 we are now COLs video engineering group
cygmatic 20:09 its not been that long and now my project is basically already legacy :)
taxilian 20:09 lol
yeah; that's how it goes
aren't you glad your ideas went into something that will get used?
cygmatic 20:09 oh yes
taxilian 20:09 and when I write my VM this semester, I think I'm going to use your functor/bind method from JSAPIAuto to make it really easy to add "assembly" calls :-P
a little extra work on the front and adding things on should be really easy
cygmatic 20:09 but the other project will still get used, its only down to bug fixing now
hm, vm sounds like fun
taxilian 20:09 it should be
it's just that it takes a lot of time and concentration :-P
speaking of which, I better finish writing this paper… :/
cygmatic 20:09 good writing then i guess
taxilian 20:09 oh
another thing we need to find soon
is a logo
something to think about
cygmatic 20:09 i assume you don't like the idea of a burning ie symbol?
taxilian 20:09 lol
but not quite the idea we're looking for
but maybe some combination (each taking 1/4 of the circle) of the logos for Safari, Chrome, IE, and Firefox...
and the resulting globe either burning or sheathed in a flame...
cygmatic 20:09 hm, i think thats too much on details
taxilian 20:09 maybe yes, maybe no; the four-in-one circle logo would be easy enough, I think
doesn't have to be precise
and a simple flame wouldn't have to be much extra
I dunno
I'd have to look at an example
I can kinda picture it, but I could never produce it
cygmatic 20:09 oh well, i don't want to sound overly negative again
taxilian 20:09 hehe
no prob
I'm open to anything
okay, really going to focus on the paper now :-P
cygmatic 20:09 good luck
taxilian 20:09 thx
taxilian 21:09 wow. we got over 200 hits today
that's more than any day in the last 6 months except last thursday and when ajaxian blogged about us
cygmatic 21:09 might be due to me answering some old answer on SO and the help forum entry on ohloh
taxilian 21:09 could be
hmm. 46% from google, 17% direct, 9% from SO
then freshmeat
actually, I bet the release hit freshmeat today
cygmatic 21:09 oh, ok
taxilian 21:09 interestingly enough, releasing often is probably the best way to get new people there
because there are sites like freshmeat that will feature you every time you have a release
and ohloh probably does some as well
cygmatic 21:09 heh, a link search on google also turns up interesting things like ... cyrillic writing of "plugin" that x-chat doesn't want to paste here
taxilian 21:09 actually, for good reason
that site has had at least one post about us
possibly more
I don't monitor it real carefully
they translated the google code front page into Russian and posted it
with a link
the random uses of a second language =]
cygmatic 21:09 hm, but that looks like a real short post with mentioning the w3c-api plugin
taxilian 21:09 possible; not sure which page specifically you were pointing to
we really need to update that plugin to 1.2.1 and submit the fix
I started, but haven't had time
it's actually pretty well written