02:10 <sb0__> I take it back. I wasn't using setjmp() correctly.

05:20 <GitHub46> [ARTIQ] sbourdeauducq pushed 6 new commits to master: http://git.io/GJg8Ww

05:20 <GitHub46> ARTIQ/master ec7a929 Sebastien Bourdeauducq: soc/runtime: provide exception handling services

05:20 <GitHub46> ARTIQ/master a8df170 Sebastien Bourdeauducq: py2llvm: support raise statement

05:20 <GitHub46> ARTIQ/master b37ceb3 Sebastien Bourdeauducq: soc/runtime: fix use of setjmp

05:54 <sb0__> whee! we can throw exceptions from the device

06:56 <mithro> have you guys had any experience with Lattice products?

07:01 <sb0__> the fpgas?

07:01 <mithro> sb0__: yeah

07:02 <sb0__> tried their software one... worse than ISE... gave up

07:02 <sb0__> *once

07:02 <mithro> sb0__: yeah, that is my understanding of most people's experiances

07:02 <mithro> sb0__: you guys generally use Altium for your designs right?

07:03 <sb0__> the m1 and mixxeo mainboards are designed with altium, yes

07:03 <sb0__> I don't like altium very much, but kicad and geda are worse...

07:04 <sb0__> the kicad UI is full of frustrating contradictions

07:06 <sb0__> I don't know how much of that mess the CERN folks have cleaned up lately

07:07 <mithro> sb0__: Last time I used Altium, it also had quite a few frustrating contradictions and similar problems. I feel like it's pretty much common to any software which has "CAD" in it's name :/

07:08 <sb0__> yeah, altium is not perfect - that and it's proprietary, windows, and bloated

07:14 <sb0__> have you tried http://circuits.io ?

07:15 <sb0__> unfortunately full of startup/cloud/maker BS and recently bought by autodesk, but it looks interesting

07:17 <sb0__> (the interface does)

15:43 <GitHub172> [migen] sbourdeauducq pushed 1 new commit to master: http://git.io/sla7ew

15:43 <GitHub172> migen/master 967b73b Florent Kermarrec: actorlib/structuring: add reverse parameter to Unpack and Pack

15:45 <sb0_> _florent_, how about removing the 'payload' subrecord completely, instead of hiding it with a getattr hack?

15:46 <GitHub143> [migen] sbourdeauducq pushed 1 new commit to master: http://git.io/QR7DdQ

15:46 <GitHub143> migen/master 66054af Florent Kermarrec: flow/actor: add packetized parameter for Sink and Source

15:52 <sb0_> also, what are the actors that cannot be used with the connect method? eventually, what the DFG does is use the same connect method...

16:00 <_florent_> hi

16:01 <_florent_> just for info, there is an error on the last patch: eop is DIR_S_TO_M too, sorry for that

16:03 <_florent_> payload can be removed, for some operation we will just have to do the distinction between the stb, ack, sop, eop and the others signals that are the payload.

16:04 <sb0_> ok, can you send new patches?

16:05 <_florent_> about the payload? if so, yes I will work on that

16:05 <sb0_> yes, and eop direction

16:13 <sb0_> hu, why add this Pipeline module? g = DataFlowGraph(), g.add_pipeline(m1, ... mn) does it

16:15 <sb0_> maybe we can get rid of the CompositeActor and e.g. make DataFlowGraph (or its replacement) a module itself...

16:16 <_florent_> I don't like the fact that the DataFlowGraph add the modules to the submodules, and that AutoCSR do not work on modules in the DataFlowGraph

16:16 <GitHub104> [migen] sbourdeauducq pushed 1 new commit to master: http://git.io/emsHhA

16:16 <GitHub104> migen/master a03570c Florent Kermarrec: flow/actor: fix eop direction

16:16 <_florent_> yes if DataFlowGraph is a module we won't have this behaviour

16:17 <sb0_> yeah. and we can add a 'noprefix' feature to the csr mechanism so that the name of the DFG does not appear

16:22 <_florent_> yes and thus DataFlowGraph will only connect modules and eventually insert plumbing, etc...

16:24 <sb0_> alternatively, we can make DFG not add non-abstract submodules to itself, and only connect them

16:25 <sb0_> but then it's easy to forget to add some actor you passed to DFG to the submodules, and it's more 'boilerplate' code

16:26 <_florent_> yes but the rules is the same than for the others modules, if you have in mind that DFG does only the connection you will not forget it

16:32 <_florent_> I have to go, I'll try to look at that soon, bye