00:23 <cr1901_modern> sb0 doesn't care for forks/pull requests for some reason

09:14 <attie> how do I make verilog export automatically create all the mem.init files instead of just putting everything on stdout?

09:16 <sb0_> verilog.convert(...).write(filename)

09:16 <sb0_> verilog.convert actually returns a ConvOutput object

09:16 <sb0_> not a string

15:11 <sb0> rjo, what is the purpose of making CSRConstant derive from fhdl.Constant?

15:13 <sb0> why not just use a value field that can take any python type? that would be consistent with integration.cpu_interface that supports string constants as well

15:22 <sb0> i think integration.config doesn't need to import * from migen, and integration.soc_core doesn't need to import warnings

15:25 <GitHub104> [artiq] whitequark pushed 2 new commits to new-py2llvm: http://git.io/v4yi4

15:25 <GitHub104> artiq/new-py2llvm 58db347 whitequark: transforms.iodelay_estimator: fix uninitialized access.

15:25 <GitHub104> artiq/new-py2llvm 2543daa whitequark: transforms.artiq_ir_generator: don't emit delay instruction for zero delay....

15:31 <GitHub76> [artiq] whitequark pushed 2 new commits to new-py2llvm: http://git.io/v4yXD

15:31 <GitHub76> artiq/new-py2llvm a04d0f8 whitequark: lit-test: fix inferencer/error_class test (broken in b0c6b70).

15:31 <GitHub76> artiq/new-py2llvm b9bb5fb whitequark: lit-test: fix iodelay/class test (broken in 506725f).

15:39 <felix_> sb0: how should i submit the patch for the artiq installation documentation? i'm a little confused about what the way to go would be. (submitting patches via email is slightly inconvenient for me, but i could do that too)

15:39 <sb0> what's wrong with sending patches by email?

15:40 <whitequark> can't you just do a github PR?

15:40 <whitequark> sb0: I have no idea how to make git send-patch work with my email setup, or even how is that command properly called

15:41 <whitequark> and making it generate a .eml file and then sending it is very awkward

15:41 <whitequark> git send-email? dunno

15:44 <GitHub150> [artiq] whitequark pushed 1 new commit to new-py2llvm: http://git.io/v4yHN

15:44 <GitHub150> artiq/new-py2llvm 9639a83 whitequark: transforms.artiq_ir_generator: correctly emit IfExpT with control flow....

15:46 <sb0> whitequark, every github PR results in a slightly annoying merge commit, and doesn't permit trivial edits of the patch (typos etc.) before merging

15:48 <felix_> well, sending a patch as an email would involve generating the mail on one machine, but sending it from another. but if thats the preferred way, i'll do that

15:48 <whitequark> well, maybe, but I don't blame people for not being arsed to send a patch by email, because screw that

15:49 <whitequark> as for the typos, it's the responsibility of the PR author to fix it and rebase...

16:03 <GitHub121> [artiq] whitequark pushed 2 new commits to new-py2llvm: http://git.io/v4yp0

16:03 <GitHub121> artiq/new-py2llvm 025bfbe whitequark: transforms.llvm_ir_generator: accept delay instructions....

16:03 <GitHub121> artiq/new-py2llvm 00ec574 whitequark: transforms.interleaver: implement (without inlining).

16:03 <whitequark> sb0: told you, i'll do it today

16:20 <sb0> rjo, how large is your cordic core again?

16:20 <sb0> in luts/dsp blocks

17:52 <rjo> sb0: a handful luts if you only need a bit or two ;)

17:52 <rjo> sb0: always zero dsp blocks

17:58 <rjo> sb0: i am not aware of strings being used as constants. are they really needed and useful? if yes, what about arrays, e.g. coefficient tables?

17:59 <rjo> sb0: I made CSRConstant a Constant because it is more concise.

18:05 <rjo> sb0: i think it is some 80 slices for a 16 bit iterative cordic, about 200 for pipelined or combinatorial. that is with a bit if IO included.