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
<GitHub95>
[smoltcp] dlrobertson commented on pull request #72 5682ccf: Good question. Would the conversions typically happen in `Repr::lower` https://git.io/vFgpS
<sb0>
if I understand correctly, with 1000base-x, you can loopback the tx pairs into rx pairs just by connecting them together, and the link should establish and transmitted packets should be received?
<_florent_>
sb0: for interfame gap, this needs to be tested. To be sure you could modify the gap module to handle the ack (for the gap to always be >= minimum gap)
rohitksingh_work has quit [Read error: Connection reset by peer]
rohitksingh has joined #m-labs
FabM has quit [Quit: ChatZilla 0.9.93 [Firefox 52.3.0/20170811091919]]
<cr1901_modern>
rjo: I have two other PRs to make right now (both platforms). One of them depends on the icestorm fixes, but is otherwise ready. Would you object if I made a PR anyway?
<cr1901_modern>
(My changes are purely additive and won't conflict w/ the icestorm improvements).
<rjo>
cr1901_modern: sure go ahead. in any case, i don't test any of that: you should write unittests. we do run those.
<cr1901_modern>
rjo: Tests for what? (I tested that the platform works on my own machine, of course)
<rjo>
cr1901_modern: even a test that makes sure the platform and the toolchain don't break would be useful.
<cr1901_modern>
That's gonna make travis builds slower b/c it needs to grab yosys and friends, just fyi
<cr1901_modern>
s/travis/buildbot
<cr1901_modern>
where is my mind today
<cr1901_modern>
rjo: "we do run those." This refers to the buildbot, correct?
<rjo>
don't run yosys. just write verilog. there are lots of ways that can break already.
<cr1901_modern>
Ahh okay, so basically a contrived module that emits verilog for all the currently-supported lattice platforms would be a decent test?
<rjo>
doesn't need to be all platforms, just cover as much code as easily possible.