| *** tpb has joined #symbiflow | 00:00 | |
| FFY00 | didn't know, thanks :D | 00:00 |
|---|---|---|
| FFY00 | mithro, btw I am also interested in packaging the symbiflow toolchains | 00:02 |
| FFY00 | but I think I am going to wait for a nextpnr release | 00:02 |
| FFY00 | which itself is waiting on yosys | 00:03 |
| mithro | FFY00: My experience with distro packaging has been that someone does the packaging once and then the original packager gets busy and it ends up becoming super obsolete quickly | 00:05 |
| FFY00 | mithro, which distros were those? | 00:08 |
| FFY00 | by experience with debian based distros has not been great | 00:09 |
| mithro | FFY00: everything were I haven't also been the package maintainer or the distro doesn't package directly and automatically from git (Ubuntu, Debian, Fedora, SuSE, slackware, etc) | 00:09 |
| FFY00 | IMO they overcomplicate packaging too much | 00:09 |
| FFY00 | in arch things are really simple | 00:10 |
| FFY00 | let's take nmigen as an example: https://git.archlinux.org/svntogit/community.git/tree/trunk/PKGBUILD?h=packages/python-nmigen | 00:10 |
| tpb | Title: PKGBUILD « trunk - svntogit/community.git - Git clone of the 'community' repository (at git.archlinux.org) | 00:10 |
| FFY00 | that's all you need to build it | 00:10 |
| FFY00 | when it updates, just bump the pkgver and update the checksums (which can be automated) and build | 00:11 |
| FFY00 | it's literally 2 commands in my terminal | 00:12 |
| mithro | FFY00: The packaging is never the problem -- the issue is keeping the packages rolling to the new versions | 00:13 |
| FFY00 | but when the packaging itself is trivial it makes things work much better | 00:14 |
| FFY00 | at least this has been my experience | 00:14 |
| mithro | FFY00: Generally what happens if that people get busy or move on to other things | 00:15 |
| FFY00 | in arch, I might ask someone else to co-maintain the packages | 00:15 |
| FFY00 | or to take over | 00:15 |
| mithro | FFY00: Places where packaging has worked okay is when there is a group of people assisted with automated systems | 00:16 |
| FFY00 | if I am busy I can just hop on the irc channel and ask for people to update my packages | 00:16 |
| FFY00 | in arch the building is automated | 00:17 |
| FFY00 | but you still need to manually update it | 00:17 |
| mithro | FFY00: Debian has gotten a lot better now they have tooling that monitors upstream for releases and automatically creates bugs and nag emails when the package is behind | 00:17 |
| FFY00 | by this I mean: it won't update automatically on an upstream release | 00:17 |
| FFY00 | each packager in arch as its own way to monitor upstreams | 00:19 |
| FFY00 | most use nvchecker | 00:19 |
| FFY00 | but it works pretty well | 00:20 |
| FFY00 | mithro, I hope that you'll have a good experience with arch :) | 00:26 |
| FFY00 | I'll try to make it happen | 00:26 |
| mithro | FFY00: I would say that fpga toolchains and getting good cross compiler packages make more sense for packaging then the litex | 00:30 |
| FFY00 | we already have yosys | 00:32 |
| FFY00 | waiting on nextpnr | 00:32 |
| FFY00 | what else should I be looking for? | 00:32 |
| mithro | FFY00: https://docs.google.com/document/d/1fWV259IbgnenwjRswE9dqCMBfyg8NC2Ily3pGQcw5ck/edit# | 00:33 |
| tpb | Title: SymbiFlow, LiteX & TimVideos Conda Packages - Google Docs (at docs.google.com) | 00:33 |
| FFY00 | we have arm, aarch64 and riscv64 cross-compilers | 00:34 |
| FFY00 | I also maintain a or1k toolchain in my repo | 00:34 |
| FFY00 | but I can move to the official ones | 00:35 |
| FFY00 | since I have the same work | 00:35 |
| FFY00 | actually is easier to maintain them there | 00:35 |
| mithro | FFY00: note there is a difference between bare metal, linux, newlib and musl cross compilers | 00:35 |
| FFY00 | ah, right forgot some were linux | 00:36 |
| FFY00 | IIRC arm is the only bare metal one | 00:36 |
| FFY00 | my or1k is also bare metal | 00:36 |
| mithro | FFY00: There is a priority number assigned to things too | 00:37 |
| FFY00 | both of them are built on newlib | 00:37 |
| mithro | There are also a *lot* of broken RISC-V cross compilers out there -- make sure multilib is enabled | 00:38 |
| FFY00 | okay | 00:38 |
| FFY00 | bare metal with no C lib is the one we use for bootstrapping right? | 00:39 |
| FFY00 | but why do we need to have a separate compiler without libc? | 00:43 |
| FFY00 | can't you just use -nostdlib? | 00:44 |
| FFY00 | also, any reason you are still on gcc 8? | 00:51 |
| FFY00 | or just time | 00:51 |
| *** OmniMancer1 has joined #symbiflow | 01:25 | |
| *** Degi has quit IRC | 02:01 | |
| *** Degi has joined #symbiflow | 02:03 | |
| *** citypw has joined #symbiflow | 02:31 | |
| *** futarisIRCcloud has quit IRC | 02:32 | |
| *** tcal has quit IRC | 02:40 | |
| *** kraiskil has joined #symbiflow | 07:35 | |
| *** Bertl_oO is now known as Bertl_zZ | 08:20 | |
| *** nonlinear has quit IRC | 08:21 | |
| *** nonlinear has joined #symbiflow | 08:31 | |
| *** Psomvanshi has joined #symbiflow | 10:21 | |
| *** Psomvanshi has quit IRC | 11:48 | |
| *** OmniMancer has joined #symbiflow | 12:00 | |
| *** OmniMancer1 has quit IRC | 12:01 | |
| *** az0re has quit IRC | 13:43 | |
| *** Bertl_zZ is now known as Bertl | 15:13 | |
| *** DraGonBusterz has quit IRC | 15:22 | |
| *** proteus-dude has joined #symbiflow | 15:48 | |
| *** proteus-guy has quit IRC | 15:52 | |
| *** citypw has quit IRC | 16:15 | |
| *** OmniMancer has quit IRC | 16:43 | |
| *** az0re has joined #symbiflow | 17:07 | |
| *** josi9824 has joined #symbiflow | 17:18 | |
| *** Bertl is now known as Bertl_oO | 18:30 | |
| *** az0re has quit IRC | 19:03 | |
| *** Vonter has quit IRC | 19:28 | |
| *** Vonter has joined #symbiflow | 19:33 | |
| FFY00 | mithro, I pushed gcc for riscv32, riscv64 and or1k elf newlib | 19:59 |
| FFY00 | currently compiling lm32 | 20:00 |
| FFY00 | probably tomorrow gonna do linux musl | 20:00 |
| FFY00 | lm32 is done | 20:16 |
| mithro | FFY00: I'm not a cross compiler expert -- I think it has something to do with calling conventions or something | 20:26 |
| FFY00 | the calling convention is defined by the abi | 20:28 |
| FFY00 | which is the same for both | 20:28 |
| mithro | FFY00: Some architectures have different ABIs for bare metal verse libc based systems | 20:29 |
| FFY00 | okay | 20:29 |
| FFY00 | ah, I think arm does | 20:29 |
| FFY00 | but most don't | 20:30 |
| mithro | FFY00: and RISC-V has a couple of ABI options I believe too | 20:30 |
| FFY00 | hum, I couldn't find anything from a quick search | 20:31 |
| mithro | FFY00: The RV32 ABI in Linux isn't stable yet either if I understand correctly | 20:31 |
| FFY00 | okay | 20:32 |
| FFY00 | well, I ordered a epc5 board | 20:34 |
| FFY00 | I can test when it arrives | 20:34 |
| *** tcal has joined #symbiflow | 20:57 | |
| *** acomodi has joined #symbiflow | 21:23 | |
| *** Harsimrat has joined #symbiflow | 21:50 | |
| *** Harsimrat has quit IRC | 21:54 | |
| *** kraiskil has quit IRC | 22:48 | |
| *** y2kbugger has joined #symbiflow | 23:26 | |
| *** acomodi has quit IRC | 23:33 | |
| *** y2kbugger_ has joined #symbiflow | 23:34 | |
| *** y2kbugger_ has quit IRC | 23:35 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!