IRC Log Viewer » #firebreath » 2017-01-24

IRC Nick Time (GMT-7) Message
Alex_SG 21:01 good afternoon all
taxilian 21:01 'evening
Alex_SG 21:01 it looks like we re making progress on 2.0 :)
taxilian 21:01 hehe. the problem is that I am not able to spend much time on it anymore so it tends to happen in spurts :-/
Alex_SG 21:01 don t worry about it
taxilian 21:01 right now I'm trying to figure out the vs2015 issue
Alex_SG 21:01 I have the same problem
sergio and serbi and a couple of others are on my payroll to address that.
taxilian 21:01 well, I'm certainly happy to help as I can with that. FireBreath 2 was never completely nailed down because frankly I ran out of time and nobody was helping
with a little help I think we could get it to the point that it's really not that hard to use
Alex_SG 21:01 I will commit time and ressources.
i think we are not far away from that point.
taxilian 21:01 I will do my best; unfortunately I'm a week behind on this (it took a ridiculously long time to get my dev environment set up what with having to install 5 versions of vs....) and there are 3 other completely unrelated projects that urgently need my attention
Alex_SG 21:01 don t worry, we have no specific deadline.
taxilian 21:01 but I'm hoping if I can nail down the few things that I'm uniquely able to deal with then others can help with other things
Alex_SG 21:01 we ll adapt to your speed.
or go in the code and hack ourselves :D
swhich should be slower :)
but since it s a long term commitment, I see that as an investement
taxilian 21:01 I am usually in this channel during business hours; sometimes I don't notice activity is all, and usually people leave before I get back to answer
Alex_SG 21:01 once we have things more stable, we will also help on the documentation
ok. But my working hours are singapore hours
so ...
we might not overlap that much, that s why e-mail (offline async) is also nice sometimes :)
for today,if you just push Boost 1.63 it will be a gr4eat help already
taxilian 21:01 hehe. true. I guess we could try setting up slack for FireBreath
I have pushed it now
Alex_SG 21:01 then I will do the tests with Win 32 bits, both 2013 and 2015
taxilian 21:01 I can also tell you how to do the update if you want to document it for me
it's not really automated, but it's also not all that hard anymore
Alex_SG 21:01 yes, if you teach me, i ll document it, so you don t waste your time twice
taxilian 21:01 basically the process is:
1. Download the version of boost you want
2. Extract boost
3. Build the tools so you have bcp.exe
Alex_SG 21:01 ok
taxilian 21:01 4. see for the command you need to use to extract the parts of boost that we use into another directory
5. execute in that directory to delete all the unneeded files -- depending on the version there might be things this misses. Basically you want to keep only the things in libs/ that are required to actually build boost
6. compare the libs/* directories on the old vs new versions to see if the file structure is the same; do it one directory at a time and copy CMakeLists.txt from each directory to the new one. Update the CMakeLists.txt file if needed (I did not need to from 1.57.0 to 1.63.0)
7. Edit boost/config/auto_link.hpp and put a #if 0 at the top and #endif at the bottom -- this prevents boost from trying to auto link when using our subset, since we'll include it with cmake. see
8. Create a new branch and commit the files to the firebreath-boost repository
Alex_SG 21:01 ok noted
taxilian 22:01 you know how to add documentation to the website?
Alex_SG 22:01 it s a wiki, right, i just need to login
i think i did it once long t ime ago
taxilian 22:01 it's running confluence, so it's basically a wiki yes
Alex_SG 22:01 did you push boost ?
taxilian 22:01 yes
Alex_SG 22:01 trying
ok, appeyor builds queued
ah! forgot to update the appveyor script to use the 32 bits msvc smake generator
ok, will do it next.
taxilian 22:01 cool
sigh. wish my new computer were here
Alex_SG 22:01 :)
taxilian 22:01 so the vs2015 thing has to do with the variantPromise type specifically; something about the template I think it doesn't like
Alex_SG 22:01 that s what sergio said
I can t follow at that level
that why i was trying to get yo uinvolved
to validate the proposed solution
taxilian 22:01 heh. well, heavily templated c++ is definitely not trivial
Alex_SG 22:01 oh, i did some
i used to write meta templates
in ITK
it s just, not in FB, so it takes time to get the arch.
taxilian 22:01 interestingly enough, it does not seem to affect other types of FB::Promise classes; only <FB::variant> specializations
Alex_SG 22:01 i rmember the joy of Traits .. 30 lines of typedef before doing anything :D
taxilian 22:01 heh. then you know how "fun" it is =]
Alex_SG 22:01 it s basically un-debuggable with noraml tools
taxilian 22:01 so I think what is actually happening is probably a conflict between implicit conversion copy constructors of FB::variant which comes into play when using it with Promise
... which probably means it's going to be a beast to track down
Alex_SG 22:01 yes, but segio proposed a way to make it compile, is what he proposed not the right way?
if you are right, you would need to surcharge the copy constructors, and be done, right?
taxilian 22:01 I can't find his PR; do you have a link?
Alex_SG 22:01 let me check if he did one
taxilian 22:01 probably something like that; I'm still trying to think through it all
maybe it was just a link, I forget
Alex_SG 22:01
taxilian 22:01 hmm. this I don't remember seeing, for some reason. I'll have to look at it
ahh, right... he initially proposed something different
Alex_SG 22:01 yes
this is is second more advanced attempt
taxilian 22:01 interesting; so the theory here is that the order it's defined in is somehow causing the issue
Alex_SG 22:01 yes, i think he just moved things around
so he was just waiting on your evaluation
taxilian 22:01 hmm, he also changed all the spaces to tabs :-P
Alex_SG 22:01 ouch
he is not a windows guy, so he did not do it in purpose
taxilian 22:01 well, not all... which makes it a bit worse :-P I'm sure he didn't
Alex_SG 22:01 on msvc 2015, appveyor complain that we are not passing /bigobj to the compiler
fatal error C1128
taxilian 22:01 huh; that's odd
it just built for me just fine
Alex_SG 22:01 it also just builds on my laptops
taxilian 22:01 I think it should probably work. I'm not really happy with the idea of moving those types into JSAPIImpl, but there isn't a technical reason that I can think of why it's a problem. I want to look at it a bit more tomorrow and make sure I've considered the implications, but I think this will probably be fine.
Alex_SG 22:01 ok
take your time
taxilian 22:01 for now, I was up way too late last night and it's already 2230 here, so I'd better go to bed. Tomorrow I'll be focussing on updating the installer templates to install for native messaging on mozilla
Alex_SG 22:01 ok. we re not workign at all on native messagign though. so we won t be able to help too much there.
taxilian 22:01 that's fine. With you guys I mainly wanted to get you unblocked, which I hope we've done =]
I on the other hand do need native messaging working -- it's my primary use of firebreath 2.0 -- so it's driving my business needs
Alex_SG 22:01 understood
taxilian 22:01 have a good.... day, I assume? =] I'm headed to bed. good luck