IRC logs for #openrisc Tuesday, 2014-03-18

--- Log opened Tue Mar 18 00:00:25 2014
olofk__franck__: I LOVE colors! Just bring it on07:05
olofk_And I agree. The output is really messy07:05
olofk__franck__: I just saw your patch. I did some stuff locally yesterday and also decided to put a python implementation of which in :)07:06
olofk_But I stole a 'which' implementation from the twisted framework after someone recommended it on stack overflow. Otherwise your patch is pretty much identical to what I had in mind :)07:07
_franck_web_olofk_: I also stole my "which" :)07:55
stekernI've only stolen witches08:03
stekernand, can the text blink as well, to give that web 1.0 feel to it ;)08:03
_franck_web_I could wrote something to have a scrolling as well :)08:04
stekernoh lord... I shouldn't have held off my changes to or1200-generic13:32
_franck_web_First come, first served :)13:46
stekern_franck__: isn't parseArgs broken now?13:54
stekernis what I'm speaking about particularly13:55
stekernwhy is i increased twice there?13:56
stekernand here:
stekernwhy is i increased after the strtod(argv[i], NULL)?13:56
stekernah, you changed the if-else, but it still looks broken...14:01
_franck_web_men I can't tell. It was supposed to work14:02
stekernIMO, if-continue is harder to follow than if-else in that piece of code14:07
stekernis it even compiling? there's a missing } here, no?
stekernah, no, sorry14:09
stekernit was just bad coding style ;)14:09
stekernI never omit the {} on the else if the if have them, so that threw me off14:10
_franck_web_I'm sure if you insist like this you'll find something ;)14:12
stekernthat is not right I think14:12
stekernargv will hold the string "--timeout" when the strtod is done on it14:13
stekernI'm changing other things in the parser, I'll fix it while I'm at it14:13
stekernfirst come, first served ;)14:14
stekernwasn't that what you said =)14:14
_franck_web_you see you are serving me a code fix14:15
stekernI more meant, I came first to see it, thus I was served with the task to fix it =)14:16
_franck_web_anyway, my bad, should have been more careful14:18
_franck_web_yeah we can also see it like this14:19
stekernlet's see how many bugs you can find back at me ;)14:58
_franck_web_:) I'll try this tonight15:00
olofk_Good to see that the cpp testbench gets some more love15:13
olofk_That is a mess15:13
_franck_web_olofk: could you try to add a dummy C src file in [verilator] section of or1200-generic and tell me if it works ?16:01
stekernolofk: yes, but as this first step I just wanted to push out what I had lying around, before it bitrots even further16:27
stekernas a follow-up, I want to do a mor1kx-generic system and move out the generic code16:28
wkoszekstekern: what's your e-mail?19:44
wkoszekjuliusb_: Ping.19:44
stekernwkoszek: stefan.kristiansson@saunalahti.fi20:04
_franck__stekern: I found a bug in parseArgs :) see github20:52
stekern_franck__: tbh, do you ever care about setting the vcd filename?21:02
_franck__no :)21:02
stekernme neither ;)21:02
stekernthis is what I have in the eco32 tb
_franck__much easier to read21:05
stekernbut, otherwise, your incremental patch for the vcd file name is good21:09
stekernso... how does one get cpp files compiled with [verilator] src_files?21:18
_franck__each core gets its src_files compiled and an archive is generated21:23
stekernbut it tries to use gcc21:23
_franck__then archives are exported in the build dir and linked together with the testbench21:24
_franck__I didn't say it works :)21:24
_franck__that's the theory21:24
stekernwhen I want it to use g++21:24
_franck__it is C (gcc) or systemC (g++)21:25
_franck__we could add a third option to source_type 'CPP'21:27
_franck__see section.py21:27
stekernso this was probably the reason behind all the headscratching about detecting .c and .cpp files21:31
stekernwould it be crazy to have c_source_files and cpp_source_files?21:32
stekernand source_files fall back to c_source_files21:32
_franck__why not, sounds good21:33
stekern_franck__: ok, here's my incremental patch with your suggestions incorporated:
stekern+ some more21:39
_franck__great. You are right about moving common parts of tb.cpp to a core21:41
_franck__we could implement a better (generic) parser :/21:42
stekernyes, that's what I'm working on21:42
stekernand I figured it'd make more sense to do the utils package as a c++ class21:43
stekerncan you somehow just "reopen" a pull-request21:46
_franck__I think if your commit has the same hash it knd of reopen it21:47
olofkKeep the pull requests coming. I have also thought about a generic parser. Is there a semi-standard way to do that in c++=21:50
olofkAnd c_source_files/cpp_source_files could work21:50
olofkBut I'm also considering a mode where the core basically says, "Hey, it's so complex to compile my stuff, so I got my own Makefile that you can run to produce the .a"21:51
_franck__olofk: can you tell me how to impoer Source in ? when I do "from fusesoc.simulator.verilator import Source" it mess things up21:56
olofk_franck__: I think I noticed something similar22:03
olofkAnd also, it looks like c files in top-level core doesn't work :(22:06
_franck__olofk: I think this make it work:
olofkThanks. I'll try that22:08
stekernolofk: I think we should use getopt22:08
olofkI tried your import thing too. It's an extremely confusing error message22:08
olofkstekern: Sounds good. I want to let cores register arguments and let the top-level create a parser from all of them22:11
olofkDoes that make sense? :)22:11
_franck__olofk: (pastie) that's what I have22:11
stekernyes, that's kind of what I had in mind22:12
olofkstekern: Perfect22:15
stekernooops, now I killed the "history" of the "old" pull-request22:15
stekernit's a bit silly this github pull-request workflow, me thinks22:16
olofk_franck__: Your pastie patch works for me22:17
olofkstekern: Got any better ideas? I like the visibility of the pull requests22:17
stekernit should somehow record the history, so you could say "okidoki, far points, I'll fix and repush to this branch"22:17
olofkah. yes. That would be nicer22:18
_franck__stekern: yeah I noticed this too22:18 I just did, and then all _franck__s comments went away...22:18
olofkI never even got to see it :)22:18
olofkDamn, I'm slow22:18
_franck__or we could use the old way and send email with patchs to the mailling list with git send-email22:19
stekernthe pull-request is there, with the stuff _franck__ commented on fixed..22:19
olofk_franck__: I'm pushing my version of your improve_verilator_root patch together with the pastie patch you sent me.22:32
olofkIs there a git way to set us both as authors?22:32
_franck__just add my signedoff if you want22:33
olofkCool. I'll do that22:33
_franck__next thing I'll do is move verilog and vpi to Are you ok for that ?22:34
_franck__then I'll add vhdl22:34
olofk_franck__: I'm a bit worried that will become very large, but you can add that stuff, and we can do a refactoring later on if that is needed22:36
_franck__ok great22:37
olofkGot to sleep now, but tomorrow night I'll be on a train for 14 hours, so I expect to be ultra-productive then :)22:44
--- Log closed Wed Mar 19 00:00:27 2014

Generated by 2.15.2 by Marius Gedminas - find it at!