sb0 changed the topic of #m-labs to: ARTIQ, Migen, MiSoC, Mixxeo & other M-Labs projects :: fka #milkymist :: Logs http://irclog.whitequark.org/m-labs
* whitequark stares
<whitequark> adding more buffering does in fact fix the problem, but it tickles another lwip bug, which causes a hang.
<whitequark> wonderful.
<GitHub171> [artiq] whitequark pushed 1 new commit to master: https://git.io/vX6qv
<GitHub171> artiq/master 18c3949 whitequark: runtime: disable the Nagle algorithm entirely....
<bb-m-labs> build #188 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/188
<bb-m-labs> build #380 of artiq-win64-test is complete: Warnings [warnings coveralls_upload] Build details are at http://buildbot.m-labs.hk/builders/artiq-win64-test/builds/380 blamelist: whitequark <whitequark@whitequark.org>
<whitequark> sb0: IT IS GREEN
<bb-m-labs> build #1089 of artiq is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1089
<cr1901_modern> whitequark: Apropos of your Ruby tweets... how does one make a parser for a context-sensitive or (worse!) undecideable language?
<cr1901_modern> I know C is nearly context-free and the solution is to keep a stack of valid identifiers when parsing.
<whitequark> cr1901_modern: step 1: get as much alcohol as you can afford
<cr1901_modern> Okay, we can go to step 2 AFTER I come back w/ the drinks :P
mumptai has quit [Ping timeout: 256 seconds]
mumptai has joined #m-labs
<sb0> whitequark, yay! thanks :)
<sb0> why do other lwip users not complain all the time? everything lwip looks broken.
fengling has joined #m-labs
fengling has quit [Ping timeout: 268 seconds]
mithro has quit [Ping timeout: 260 seconds]
mithro has joined #m-labs
bentley` has quit [Ping timeout: 248 seconds]
bb-m-labs has quit [Ping timeout: 260 seconds]
bb-m-labs has joined #m-labs
sandeepkr has joined #m-labs
sandeepkr_ has quit [Quit: Leaving]
<whitequark> sb0: dunno. which lwip users? :D
<whitequark> what's the next priorities btw? (I'm going to spend a few days merging solvespace commits though...)
<rjo> whitequark: i'd like #591, i can do #566, i guess #598 might be fixed now.
<whitequark> rjo: the reason I didn't do 591 is that it seems to require the phaser branch
<whitequark> so that needs a merge
<whitequark> what's your proposed solution for #566? just pin it to the build?
<whitequark> I don't think you can depend on [vc14] or something explicitly
<whitequark> #598 probably won't be fixed because I only fixed issues on device TX side
<whitequark> device RX side is probably still fuckde
<whitequark> but worth a check
<sb0> whitequark, i don't think pinning the build is necessary, the version should be enough
<rjo> whitequark: just let our qt5 depend on that icu version.
<rjo> whitequark: the merge is coming
<GitHub73> [artiq] jordens pushed 2 new commits to phaser: https://git.io/vX6PJ
<GitHub73> artiq/phaser aedb674 Robert Jordens: Merge branch 'master' into phaser...
<GitHub73> artiq/phaser d158c69 Robert Jordens: phaser: fix frequency comment
<whitequark> bb-m-labs: force build --props=branch=phaser artiq
<bb-m-labs> build forced [ETA 49m18s]
<bb-m-labs> I'll give a shout when the build finishes
<bb-m-labs> build #189 of artiq-board is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/189
<whitequark> oh
<whitequark> bb-m-labs: stop build artiq broken
<bb-m-labs> build 1090 interrupted
<bb-m-labs> build #1090 of artiq is complete: Exception [exception interrupted] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1090
<rjo> whitequark: my fault? your fault?
<whitequark> I *think* yours, I haven't touched the dds code in months
<whitequark> oh, hm
sb0 has quit [Quit: Leaving]
<whitequark> that might be related to the recent misoc changes?
<rjo> whitequark: oh i know...
<GitHub173> [artiq] jordens pushed 1 new commit to phaser: https://git.io/vX6Pz
<GitHub173> artiq/phaser 2e48250 Robert Jordens: phaser: fix DDS dummy cfg
<GitHub51> [artiq] jordens pushed 1 new commit to phaser: https://git.io/vX6Xi
<GitHub51> artiq/phaser 70a7032 Robert Jordens: phaser: use misoc cordic
<bb-m-labs> build #190 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/190
<whitequark> rjo: can you please flash that?
<whitequark> (only the gateware is needed)
<whitequark> fwiw I looked into restructuring the buildbot to skip gateware changes and such
<whitequark> but it's troublesome with our current dependency management
<whitequark> the gateware depends on the exact version of ARTIQ itself, which is in principle correct...
<rjo> whitequark: yes. there are to many factors in that interface.
<rjo> whitequark: but #591 doesn't need hardware per se. if it folds the shit out of the code i am happy.
<rjo> whitequark: but i'll flash anyway.
<whitequark> oh, I was thinking of testing the other change (kernels hanging)
<rjo> whitequark: i'll test the hanging thing. might that be fixed by !nagle?
<whitequark> unlikely but possible, yes
<whitequark> I cannot say anothing about lwip with a high degree of certainty...
<rjo> imho we can dramatically simplify the "install from source" instructions. just tell them what conda packages to install to be able to compile bitstreams/bios/runtime. if somebody is masochistic enough and wants to really "install from source" the conda recipes are authoritative enough.
bentley` has joined #m-labs
<rjo> whitequark: flashed
<rjo> whitequark: http://hastebin.com/ijerobadul.sql my fault?
<cr1901_modern> rjo: "the conda recipes are authoritative enough." This might've changed in the past few months, but that wasn't true when I last built ARTIQ ;)
<rjo> cr1901_modern: but they are. otherwise they are buggy.
<cr1901_modern> I currently can't build ARTIQ from source even if I wanted to b/c building the Rust runtime isn't supported on Windows. But I'll look into doing it again b/c I feel masochistic this week.
<whitequark> rjo: disagree
<whitequark> rjo: for example I don't use conda locally, but pip and the rest
<whitequark> it's quite hard to use conda if you want to e.g. develop llvm or rust
<whitequark> and anyhow the conda recipes differ from the install from source instructions in some ways...
<whitequark> re "your fault": sort of. this means an LLVM version mismatch
<rjo> whitequark: ack. but those are two different contexts. if you are an artiq user and want to hack artiq you'd be totally fine installing conda stuff and then an artiq working tree. in your case you are developing llvm, writing the conda recipes and duplicating that same information in the documentation where afaict few people need it.
<whitequark> rjo: few people, true, but artiq developers do
<whitequark> it's important for internal culture as well
<whitequark> there's nothing worse than company software that's impossible to install. and artiq already fares badly on that front, even with docs.
<rjo> whitequark: ack. but i would still think that installation instructions based on conda are immensely useful.
<rjo> whitequark: in the past, haven't the conda recipes been more reliable than the documentation?
<rjo> whitequark: and i do consider myself an artiq developer but haven't been using those instructions in the past.
<cr1901_modern> For the dependencies that don't change much, conda works fine even for development, in my experience
<rjo> i fully acknowledge that if you are developing e.g. llvm-or1k you wouldn't use the conda packages...
<cr1901_modern> Just need to do setup.py develop on the artiq (and perhaps migen/misoc) tree
<rjo> cr1901_modern: not if you are developing llvm-or1k
<whitequark> rjo: I agree that your suggestion is useful, I just don't think it should *replace* the existing install_from_source
<whitequark> you also have far higher tolerance for conda than me
<GitHub82> [conda-recipes] jordens pushed 1 new commit to master: https://github.com/m-labs/conda-recipes/commit/77efaef4b5e1129324de3cfbe5d494137311f80d
<GitHub82> conda-recipes/master 77efaef Robert Jordens: qt5: icu=56.1=0
<rjo> whitequark: the kernel hang ist still there.
<bb-m-labs> build #162 of conda-win64 is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/conda-win64/builds/162
<bb-m-labs> build #248 of conda-lin64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/248
<whitequark> rjo: as expected
<whitequark> that's really annoying.
<bb-m-labs> build #164 of conda-win32 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-win32/builds/164
<bb-m-labs> build #81 of conda-all is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/conda-all/builds/81
mumptai has quit [Quit: Verlassend]