Wednesday, 2019-04-24

*** tpb has joined #litex00:00
tpbTitle: Free Assembly for 5 Pieces – Only at Seeed Fusion! | Seeed Studio Blog (at
keesjI need to spend a few hours creating the correct bom file but it would be a great experience.. if they have ice4006:34
futarisIRCcloud - I couldn't see any Ice40 at their OPL.07:02
tpbTitle: Fusion Open Part Library(OPL) Service - Seeed Studio (at
futarisIRCcloud says that they can use digikey though...07:03
tpbTitle: Fusion PCB Manufacturing & Prototype PCB Assembly - Seeed Studio (at
keesjI tried multiple ways to use their open parts library with kicad but failed all times.07:04
keesj(incompatible with kicad 5 and other similar problems) (this was last year) perhaps I need to try again (artible from Jan this year)07:06
futarisIRCcloudCould probably do a S7 Mini or Ice40.07:07
futarisIRCcloudtinyfpga_ : Did Greg Davill get to work?07:08
tpbTitle: TinyFPGA-EX/board at master · tinyfpga/TinyFPGA-EX · GitHub (at
*** felix_ has quit IRC07:50
*** felix_ has joined #litex07:51
keesjthe new litex_server does not work as reliable as the previous one08:15
keesjspecially it does not matter if the server is up or not I will get the same strange error08:16
keesjValueError: not enough values to unpack (expected 5, got 1)08:17
keesjhmm it looks like the parsing of the csr is to blame the newly generated comment on top of the csr causes problems (e.g. the ##### Generated by litex)08:22
keesj is needed08:29
tpbTitle: Python: skip comment lines marked with # in csv.DictReader - Stack Overflow (at
keesj (the new header format to reproduce the problem)09:04
tpbTitle: Ubuntu Pastebin (at
_florent_keesj: sorry, i finish something and i do fix that right after09:06
keesjI have "fixed" it locally if you want I create a issue on github09:12
keesjthe class would need a little refactoring to not have to much duplication09:13
_florent_keesj: i'm looking at that, you can create a pull request yes if you want09:46
tpbTitle: tools/remote/csr_builder: allow comments in csv file and cleanup · enjoy-digital/[email protected] · GitHub (at
keesjyou beat me to it10:53
tpbTitle: Fix parsing of CSR files containing a header. · keesj/[email protected] · GitHub (at
keesjthe @staticmethod is nicer11:08
somlo_florent_: about get_*_git_revision(), doing something like "litex/litex/tools/ --cpu-type=vexriscv" used to work before, now it crashes with "fatal: not a git repository (or any of the parent directories): .git"13:18
somloI assume it now only works if you run it from within the right folder (e.g., the litex project directory)?13:18
somloand, what if one has a distro-packaged migen, and no git repo in which to run `git foo` commands (re. get_migen_git_revision(), specifically)?13:21
somlook, it's only get_migen_git_revision() that fails for me right now (distro-packaged migen in Fedora), but can we talk about codifying the assumption that litex (and migen) are being deployed as git repos?13:27
*** futarisIRCcloud has quit IRC14:22
somlo_florent_: how about ?17:01
tpbTitle: build: handle exceptional case when litex/migen not deployed as git repo by gsomlo · Pull Request #173 · enjoy-digital/litex · GitHub (at
keesjit the checking of git version not also only relevant when in "developer" mode? can this not reliably be detected?20:17
somlojust dug around through litex to see how generated_banner() is actually *used*20:26
somlobtw, _florent_, this line should be "+=", missed that in my PR:
tpbTitle: litex/ at master · enjoy-digital/litex · GitHub (at
somloit appears it's just there to add comments re. which migen and litex git commit was current when various generated .h files (and top.v) were produced20:29
keesjwon't break anything indeed20:29
keesj(from the looks of it)20:29
somloso "------" as a fallback is fine with me for now -- the interesting question IMHO (and to which I don't have a reasonable answer yet) is how I'd get the migen/litex version if/when they're not in a git repo (developer-mode or not)20:30
somloI'd really hate to go down the "rpm? deb? other?" rabbit hole, just for a bunch of comments :)20:30
somlocheesy cop-out, the packager (me, in Fedora's case) could sed -i 's/------/<version>/' as part of the package build process :)20:32
somloi.e. just monkeypatch the version into the sources. But I'm not sure I want to go down in history as the guy who suggested that :)20:33
keesjthis problem has been solved many times already and also keeps comming back. I try to keep away from it20:36
keesjtesting for the existance of the .git folder is probably way nicers of a solution20:38
somlokeesj: right, but the question is what to do if .git is *not* there (the "except:" branch)20:39
_florent_somlo: i think your solution is fine for now.  If migen/litex are installed from .rpm/.deb we'll fall back to the "--------" case, we could fix that later if really needed20:50
somlo_florent_: thanks, that works for me!20:53
_florent_somlo: have you been able to do progress on the rocket integration?20:56
_florent_somlo: are you still stuck with the missing burst support?20:56
somlo_florent_: I found a third-party axi-wb bridge here:
somlowhich I'm going to try tomorrow20:58
somloif that won't work, I'll try to start hacking on the LiteX internal axi implementation20:59
somloI may be working harder than needed just to avoid learning something :) but I really just want to see something (anything at all) work (even if horribly inefficient) before deciding where to spend serious effort optimizing things (and that's when I'll actually *have* to learn things :)21:00
_florent_ok, if you are stuck i can try to help creating the axi to wishbone bridge21:00
_florent_yes sure, i understand21:00
somloI read somewhere that if laziness were a martial art, they'd call it "the Path of Least Resistance" :)21:01
somloit's almost quittin' time here (EDT) but I'm planning to lock myself in my office tomorrow and do nothing but this, so chances are I'll be here crying about it...21:07

Generated by 2.13.1 by Marius Gedminas - find it at!