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
fengling has quit [Ping timeout: 240 seconds]
fengling has joined #m-labs
sandeepkr has joined #m-labs
<sb0> why has no one made a DIY ultracapacitor-powered Bitter electromagnet?
<cr1901_modern> Try to make one yourself?
fengling has quit [Ping timeout: 240 seconds]
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
<mithro> ysionneau: Doesn't look like we ended up with any strong students for RTEMS or Linux port to MiSoC
<mithro> You guys use clang for C code compiling for artiq?
<cr1901_modern> Yes; or1k only has a clang port I think
<mithro> cr1901_modern: There is an "out of tree" gcc
<mithro> whitequark: Why do you build the X86 version as well as the or1k version at https://github.com/m-labs/conda-recipes/blob/master/conda/llvm-or1k/build.sh ?
<whitequark> tests
<whitequark> I use JIT to run some simpler tests on the host instead of coredevice
<mithro> whitequark: so clang can be used to compile the MiSoC bios and stuff?
<whitequark> sure
<whitequark> that's how it is built for OR1K
<whitequark> I had to fix a few integrated assembler bugs
<mithro> whitequark: shouldn't llvm-or1k require binutils? Or does llvm also replace all those parts too?
<whitequark> conda install binutils-or1k
<whitequark> LLVM, unfortunately, lacks a linker
<mithro> whitequark: It's not listed as a dep in the meta.yaml (at https://github.com/m-labs/conda-recipes/blob/master/conda/llvm-or1k/meta.yaml)
<whitequark> it's not a dependency of LLVM
<whitequark> if you never need to link, LLVM is just fine
<sb0> another pyqt fuckup: using integers as qt internalPointers causes segfault
<sb0> also, QAbstractItemModel is horrible to use, as I said earlier
fengling has joined #m-labs
<sb0> not only horizontal autolayout is broken, the vertical one has a good share of bugs too
<sb0> so I'm trying to use a tree for displaying the log...
fengling has quit [Ping timeout: 240 seconds]
<sb0> hm, it's something else that is causing the segfault. but it segfaults...
<mithro> whitequark: I've added gcc-4.8 to my travis packages, but llvm build is failing with " Host GCC version must be at least 4.7!" - Any idea how to get it to point to the gcc-4.8 rather than gcc-4.6?
<mithro> whitequark: your travis config, nor conda config seem to do anything special?
<whitequark> need to change the conda recipe to pass -DCMAKE_C_COMPILER -DCMAKE_CXX_COMPILER
<whitequark> my travis config changes the default gcc to 4.8
<whitequark> i.e. the one at /usr/bin/gcc
<mithro> whitequark: Where in the travis config does it do that? I think I'm missing something?
<sb0> the reason for the segfault is pyqt is obviously unable to track the internalPointer, and the garbage collector frees the object unless you do something to keep it alive
<sb0> grrrrr
<sb0> so it's a bit like using malloc/free in C, except that in C those calls are explicit and well-controlled, but here you have to play games with the garbage collector
<sb0> yay pyqt
<whitequark> mithro: I think it's pulled via some dep
fengling has joined #m-labs
<sb0> QTreeView.header().resizeSections(QtWidgets.QHeaderView.ResizeToContents), another broken Qt function
<sb0> if you have a horizontal scrollbar enabled, it does nothing that could change the scroll range
<sb0> i.e. if you have a row with a lot of text that would require the scroll bar, calling it does nothing and the user has to resize the header manually
<sb0> which is pretty inconvenient (requires drag and drop out of the window)
acathla has quit [Quit: Coyote finally caught me]
acathla has joined #m-labs
acathla has joined #m-labs
fengling has quit [Quit: WeeChat 1.4]
<whitequark> that seems odd
<GitHub58> [artiq] sbourdeauducq pushed 3 new commits to master: https://git.io/vahij
<GitHub58> artiq/master e9ca125 Sebastien Bourdeauducq: protocols/remote_exec: fix call
<GitHub58> artiq/master b190581 Sebastien Bourdeauducq: Revert "runtime/i2c: more conservative timings"...
<GitHub58> artiq/master bebd89c Sebastien Bourdeauducq: gui: redesign table/trees to avoid slow and buggy qt/pyqt autosize. Closes #182. Closes #187.
<sb0> whitequark, see for yourself
<bb-m-labs> build #479 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/479 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
sandeepkr has quit [Ping timeout: 244 seconds]
<sb0> whitequark, any update on the compiler issues?
<whitequark> still working on it.
<GitHub0> [migen] enjoy-digital pushed 2 new commits to master: https://git.io/vahSd
<GitHub0> migen/master e45eaa4 Florent Kermarrec: fhdl/simplify/MemoryToArray: remove false we_granularity limitation (we_granularity is implemented and has been tested)
<GitHub0> migen/master a8fd434 Florent Kermarrec: fhdl/simplify/FullMemoryWE: fix typo
<GitHub179> [artiq] jordens pushed 1 new commit to master: https://git.io/vah9g
<GitHub179> artiq/master 5b536d7 Robert Jordens: i2c: fix variable name (closes #347)
<GitHub38> [artiq] jordens pushed 1 new commit to release-1: https://git.io/vah9V
<GitHub38> artiq/release-1 358d2a6 Robert Jordens: i2c: fix variable name (closes #347)
<bb-m-labs> build #480 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/480
<whitequark> sigh
<whitequark> I'll go to the lab today and fix it
<GitHub57> [artiq] sbourdeauducq pushed 5 new commits to master: https://git.io/vahHd
<GitHub57> artiq/master ca4e6e0 Sebastien Bourdeauducq: master/worker: use only first line in short_exc_info
<GitHub57> artiq/master 67ab8db Sebastien Bourdeauducq: protocols/pc_rpc: short_exc_info
<GitHub57> artiq/master 5c6e2d7 Sebastien Bourdeauducq: gui/log: send Qt model notifications correctly
<bb-m-labs> build #481 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/481 blamelist: Robert Jordens <rj@m-labs.hk>
<GitHub65> [artiq] sbourdeauducq pushed 5 new commits to release-1: https://git.io/vahQ4
<GitHub65> artiq/release-1 5d5a443 Sebastien Bourdeauducq: gui: redesign table/trees to avoid slow and buggy qt/pyqt autosize. Closes #182. Closes #187.
<GitHub65> artiq/release-1 7bdec1b Sebastien Bourdeauducq: master/worker: use only first line in short_exc_info
<GitHub65> artiq/release-1 74b3c47 Sebastien Bourdeauducq: protocols/pc_rpc: short_exc_info
<GitHub175> [artiq] sbourdeauducq pushed 1 new commit to master: https://git.io/vah7t
<GitHub175> artiq/master 4c62219 Sebastien Bourdeauducq: doc: update flterm instructions. Closes #346
<GitHub94> [artiq] sbourdeauducq pushed 1 new commit to release-1: https://git.io/vah7q
<GitHub94> artiq/release-1 027aa5d Sebastien Bourdeauducq: doc: update flterm instructions. Closes #346
<GitHub134> [artiq] jordens pushed 1 new commit to release-1: https://git.io/vah7b
<GitHub134> artiq/release-1 9de11dd Robert Jordens: RELEASE_NOTES: pipistrello speed change
<GitHub185> [artiq] jordens pushed 1 new commit to master: https://git.io/vah7N
<GitHub185> artiq/master 8a9c1d3 Robert Jordens: RELEASE_NOTES: pipistrello speed change
<bb-m-labs> build #482 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/482 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<bb-m-labs> build #483 of artiq is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/483 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<bb-m-labs> build #484 of artiq is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/484 blamelist: Robert Jordens <rj@m-labs.hk>
sandeepkr has joined #m-labs
_rht has joined #m-labs
<mithro> rjo: I found another issue which might be related to your pipistrello issue - https://github.com/m-labs/migen/issues/43
<mithro> rjo: I'm unsure how to test my fix, as I'm still stuck on the legacy branch
<whitequark> ok. i figured a good way to solve that.
<whitequark> we'll get better performance, smaller code *and* less hacks in the compiler
<rjo> mithro: what pipistrello issue?
<mithro> rjo: The one you thought was non-determinism?
<rjo> tigs and inferred constraints seem just fine the way i did them. and how would that affect determinism?
<mithro> rjo: ISE is stupid and if you have a period going into a PLL or DCM it won't replicate them through if you have a TIG constraint on the same net
<mithro> rjo: Which means that ISE wouldn't be applying the correct constraints to anything after a PLL or DCM in your design
rohitksingh has joined #m-labs
<rjo> that's completely orthogonal to inherently non-deterministic routing.
<rjo> also pretty sure that there were the correct constraints left over in our case.
<rjo> but thanks for the heads up.
rohitksingh has quit [Ping timeout: 244 seconds]
rohitksingh has joined #m-labs
sandeepkr_ has joined #m-labs
sandeepkr has quit [Read error: No route to host]
sandeepkr__ has joined #m-labs
sandeepkr_ has quit [Read error: No route to host]
sandeepkr has joined #m-labs
sandeepkr__ has quit [Read error: No route to host]
sandeepkr_ has joined #m-labs
sandeepkr has quit [Ping timeout: 264 seconds]
sandeepkr_ has quit [Ping timeout: 248 seconds]
kuldeep has quit [Remote host closed the connection]
kuldeep has joined #m-labs
_rht has quit [Quit: Connection closed for inactivity]
rohitksingh has quit [Quit: Leaving.]
<GitHub159> [artiq] whitequark pushed 3 new commits to master: https://git.io/vVeg5
<GitHub159> artiq/master 39599d4 whitequark: compiler: get rid of the GetConstructor opcode.
<GitHub159> artiq/master 39d2379 whitequark: transforms.llvm_ir_generator: extract class function attributes....
<GitHub159> artiq/master 8d05666 whitequark: compiler: quote functions directly instead of going through a local.
<whitequark> sb0: ok. mostly done. i did the compiler work i thought of yesterday.
<whitequark> i've just now realized it's not actually enough to properly solve embedding, but it will be enough to get 1.0 out of the door
<whitequark> i also broke the "smart" interleaving. to fix it i'll have to implement the parts that are missing ("not actually enough" above)
<bb-m-labs> build #485 of artiq is complete: Failure [failed lit_test] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/485 blamelist: whitequark <whitequark@whitequark.org>
<whitequark> bb-m-labs: force build --props=package=llvmlite-artiq conda-all
<bb-m-labs> build #16 forced
<bb-m-labs> I'll give a shout when the build finishes
<GitHub120> [conda-recipes] whitequark pushed 1 new commit to master: https://github.com/m-labs/conda-recipes/commit/2ee5da8478b755c27f0aa08d4fcc15115dca3814
<GitHub120> conda-recipes/master 2ee5da8 whitequark: llvmlite-artiq: bump.
<whitequark> anyway, regarding #298. it, uh, became worse: from 36us to 42us.
<whitequark> but tuning the LLVM pipeline (#315) is going to fix it, now that i've unblocked the things LLVM could not see through
<whitequark> i'll do this later today
<whitequark> the build is broken because the llvmlite fix is not in conda yet.