#milkymist IRC log for Monday, 2011-12-12

wpwrakaha. and mtk doesn't seem to like " in messages00:32
cladamw_wpwrak, How's the package ? Not sure the invoice from Wolfgang can help you or not.01:25
wpwrakcladamw_: i'll try to find time to visit customs tomorrow. then i'll see. thu and fri were holidays, so tomorrow is the first day when i can try to salvage it.01:28
wpwrak52 commits ahead :)01:30
cladamw_wpwrak, i see. I do really sorry that let it into this situation.01:30
wpwrakah well, let's hope the customs guys had a nice long weekend and are friendly :)01:31
GitHub151[milkymist] sbourdeauducq pushed 5 new commits to master: https://github.com/milkymist/milkymist/compare/92eefac...b92595f10:59
GitHub151[milkymist/master] fpvm.c: renaming decision in fpvm_assign got reversed - Werner Almesberger10:59
GitHub151[milkymist/master] libfpvm: add also functions and internal variables to fnp.ids - Werner Almesberger10:59
GitHub151[milkymist/master] libfpvm/unique.c: fix grow_table and strcmp_n - Werner Almesberger10:59
GitHub42[milkymist] sbourdeauducq pushed 2 new commits to master: https://github.com/milkymist/milkymist/compare/b92595f...75e8acb11:00
GitHub42[milkymist/master] libfpvm/Makefile: build now for RTEMS (instead of the demo firmware) - Werner Almesberger11:00
GitHub42[milkymist/master] libfpvm: git rm -rf lm32-rtems/ - Werner Almesberger11:00
barbu-uucphttp://www.youtube.com/watch?v=ywWBy6J5gz811:18
wpwrakoh dear. so that's what really happens in those electronic brains11:33
kristianpaullol11:33
wpwrakhmm, no trip to the airport today either. the national bank is on strike, to i couldn't pay the customs fees.11:35
kristianpaulintentional strike?11:37
kristianpauli dont read too much news afaik..11:37
wpwrakare there any unintentional strikes ? :)11:43
cladamw( 3 thinner traces between balls of spartan-6 fpga ) http://en.qi-hardware.com/wiki/File:M1rc3_0x74_XC6SLX45-2FGG484C_without_balls_2.png11:50
barbu-uucpcladamw: ?11:52
barbu-uucpyeah, many BGA packages are based on a piece of multilayer PCB with the chip bonded on it11:52
cladamwyes. amazing though.11:54
barbu-uucpyou can take an old motherboard and crack open the BGA packages11:54
barbu-uucp(separate the PCB from its cover)11:54
barbu-uucpyou'll see the chip and everything11:54
cladamwhe. That fpga I'll still use it. :-)11:58
GitHub152[milkymist] sbourdeauducq pushed 4 new commits to master: https://github.com/milkymist/milkymist/compare/75e8acb...4fdae4a12:09
GitHub152[milkymist/master] libfpvm: don't use TOK_* inside fpvm.c - Werner Almesberger12:09
GitHub152[milkymist/master] libfpvm: split parser from fpvm.c (to fpvm-export.c) - Werner Almesberger12:09
GitHub152[milkymist/master] libfpvm: export ast.h and split internal communication (to parser_itf.h) - Werner Almesberger12:09
wpwrakhmm, should be fun to enable -Wshadow. produces lots of warnings :)12:11
wpwrakof course, including the annoying y0, y112:12
ThihiSo who develops the gui for MilkyMist One?12:15
ThihiAnd who develops the flickernoise language?12:15
wpwraksebastien is the principal author of everything. but some others are lending a helping hand from time to time as well.12:18
ThihiOk.12:21
GitHub139[flickernoise] sbourdeauducq pushed 2 new commits to master: http://git.io/1O2Vxw12:30
GitHub139[flickernoise/master] src/Makefile: wrap long lines and add spaces around assignments - Werner Almesberger12:30
GitHub139[flickernoise/master] arrival of the FNP scanner and parser (into src/compiler/) - Werner Almesberger12:30
Thihiwpwrak, so wait, who is sebastien!? I don't know you guys by your real names :o12:31
wolfspraulSebsatien = lekernel = sb012:34
wolfspraulwpwrak = werner12:34
ThihiOk :)12:34
wolfspraulSebastien is the founder of the Milkymist project and original coder since 200712:34
wolfspraulI joined in 2010 but mostly on the manufacturing/business side (let's say 'wannabe business' :-))12:34
ThihiOk.12:35
wolfspraulthen there are lot of others like Werner, MIchael Walle, Uwe Bonnes, Takeshi Matsuya, and more12:35
wolfspraulI think in all of 2011 Werner has been taking an amazing role and lead, and now luckily it seems Sebastien is kicking back in as well :-)12:35
wolfsprauland then the new 'users', like you! :-)12:36
Thihi:)12:36
wolfsprauland kristian paul, not to forget. one of the first milkymist hackers.12:36
wolfsprauland probably more I forgot12:36
wolfspraulyeah, that's the story roughly12:36
wolfsprauloh, Lars of course12:36
wolfspraulon-and-off with the Linux port12:37
wolfspraullarsc = Lars Peter-Clausen12:37
wolfspraulthere are more, like I said. I just mention names randomly as they pop in my mind. hope nobody feels forgotten :-)12:37
Thihi:)12:40
ThihiWell, thanks for the info. Anyway, I'll prolly have some points about usability in a few weeks, so I just needed to know who to talk about them to ;P12:40
GitHub46[flickernoise] sbourdeauducq pushed 7 new commits to master: http://git.io/Qmkhxg12:41
GitHub46[flickernoise/master] src/renderer/: convert old-style function declarations - Werner Almesberger12:41
GitHub46[flickernoise/master] renderer/compiler.c: include stdbool.h - Werner Almesberger12:41
GitHub46[flickernoise/master] src/Makefile: add quiet compilation and make it the default - Werner Almesberger12:41
GitHub45[flickernoise] sbourdeauducq pushed 1 new commit to master: http://git.io/ImDgtQ12:42
GitHub45[flickernoise/master] compiler/: provide header for migrated functions fpvm_init and fpvm_assign - Werner Almesberger12:42
GitHub153[milkymist] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/milkymist/commit/7cec37ec937c12a1d9d84f61ba56e7e36903163412:43
GitHub153[milkymist/master] fpvm/fpvm.h: declare bottom functions still here, not migrated functions - Werner Almesberger12:43
GitHub64[flickernoise] sbourdeauducq pushed 5 new commits to master: http://git.io/SXEbCw13:28
GitHub64[flickernoise/master] compiler/: introduce parser start state - Werner Almesberger13:28
GitHub64[flickernoise/master] compiler: directly process assignments - Werner Almesberger13:28
GitHub64[flickernoise/master] compiler: make init_pvv use assignments in FNP notation, as a proof of concept - Werner Almesberger13:28
barbu-uucpwpwrak: yeah, original milkdrop presets often comes with syntax errors and other mistakes ...13:29
barbu-uucpyour fixes look good13:30
wpwrakcool, thanks for checking ! if you run them, you'll find that the altars of madness now look a bit more interesting13:53
barbu-uucphaha, expect :)13:54
barbu-uucplast time I heard of this tool was in 2000 or so :)13:54
wpwrakit can still be useful :)13:56
GitHub46[flickernoise] sbourdeauducq pushed 8 new commits to master: http://git.io/oVMpHQ13:58
GitHub46[flickernoise/master] compiler/ptest/: FNP parser tester - Werner Almesberger13:58
GitHub46[flickernoise/master] compiler/test/: regression test framework and a few test cases - Werner Almesberger13:58
GitHub46[flickernoise/master] patches/: fix various typos (PLEASE REVIEW !) - Werner Almesberger13:58
wpwrakbut what i call there isn't "the" expect. just my own function13:59
wpwrakit comes from ./Common, so it'll override any /usr/bin/expect that may be around14:00
barbu-uucpwpwrak: btw, RTEMS has asprintf()14:03
barbu-uucpand ptest leaks memory when error is non-NULL - it doesn't really matter here, but I'd recommend you add a free() for the example14:05
barbu-uucphmm, in fact, it's [preset00], not [preset]14:18
barbu-uucp"per-vector" variables? :)14:19
GitHub136[flickernoise] sbourdeauducq pushed 17 new commits to master: http://git.io/8veVUw14:22
GitHub136[flickernoise/master] compiler: don't feed TOK_ERROR to the parser - Werner Almesberger14:22
GitHub136[flickernoise/master] compiler: report the line number and the context of an error - Werner Almesberger14:22
GitHub136[flickernoise/master] compiler: remove trailing whitespace - Werner Almesberger14:22
GitHub74[flickernoise] sbourdeauducq pushed 2 new commits to master: http://git.io/4xjCLQ14:26
GitHub74[flickernoise/master] compiler: preset00, not preset - Sebastien Bourdeauducq14:26
GitHub74[flickernoise/master] compiler: fixed typo 'per-vector' - Sebastien Bourdeauducq14:26
GitHub142[milkymist] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/milkymist/commit/d30353b1f46a6b73fc364044ba221d964680771914:28
GitHub142[milkymist/master] milkymist: ANSI-style fpvm_version declaration - Werner Almesberger14:28
GitHub197[flickernoise] sbourdeauducq pushed 5 new commits to master: http://git.io/N9FSxA14:28
GitHub197[flickernoise/master] flickernoise: use -Wstrict-prototypes - Werner Almesberger14:28
GitHub197[flickernoise/master] flickernoise: use -Wmissing-prototypes - Werner Almesberger14:28
GitHub197[flickernoise/master] flickernoise: use -Wold-style-declaration - Werner Almesberger14:28
wpwrak(per-vector) ehm :) will be in the next set of fixes :)14:29
wpwrakah, you already fixed it14:29
wpwrakexcellent. now we're synced again. thanks a lot !14:31
barbu-uucpwpwrak: btw, there are many regressions in the patch compiler now. I guess they are expected?14:35
barbu-uucpit seems the per-vertex code isn't correct14:37
wpwrakyeah, i have just one that looks odd. but maybe with camera it would be more14:42
wpwraki need to host-build libfpvm too, then i can simply compare the generated code14:43
barbu-uucpmh, all of them look odd for me14:45
barbu-uucpthe images aren't correctly warped14:46
barbu-uucp(i.e. the internal texture buffer)14:47
barbu-uucpthis smells like wrong per-vertex code. the per-frame looks OK, though.14:47
wpwrakmaybe something in finish-pvv.fnp ...14:49
barbu-uucphmm... is libmath used anywhere except the demo fw? test tools maybe?14:56
barbu-uucpah, yes14:57
barbu-uucpsinf() is used to generate the audio test tone14:57
GitHub118[milkymist] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/milkymist/commit/6f50e9694d9cd3cdcd0227d66721305576c21ac815:03
GitHub118[milkymist/master] Remove demo firmware - Sebastien Bourdeauducq15:03
wpwrakdo things like "A Matter Of Taste" still look right for you ?15:07
barbu-uucpit won't even compile15:17
barbu-uucpfailed to add equation to per-vertex footer: failed to allocate renamed register15:17
barbu-uucpline 5715:17
barbu-uucpmaybe we're out of sync?15:17
wpwraklet's find out ...15:18
wpwrakno, looks good15:25
wpwrakdid you do a make clean on libfpvm and re-install ? just "make" may not be enough, because of the library15:26
barbu-uucpyes, I did15:29
wpwrakhmm. weird.15:29
barbu-uucpah, I still had a ./obj/renderer/compiler.o in FN15:30
barbu-uucpmaybe it linked the wrong one15:30
wpwrakthat would be interesting ;)15:30
barbu-uucpno, still won't work15:31
wpwrakvery strange. let's check the generated files15:33
wpwrakb16ab1eb337a672a2a580eb7b5240502  finish-pfv.fnp15:33
wpwrak51f7d69ac9218c41d2263cb92f68e643  finish-pvv.fnp15:33
wpwraka93ff24e5298caec3896a256a66a8c8a  init-pvv.fnp15:33
wpwrak(md5sum)15:33
wpwrakerr, nonsense15:34
wpwrak8204d891896e757b447c2367b00dc89b  infra-fnp.h15:34
wpwrakthat's the one. the others are inputs :)15:34
barbu-uucpsame md5 here15:36
barbu-uucpwant to try my flickernoise.bin?15:37
wpwrakyeah, let's swap binaries15:38
wpwrakhere is mine: http://downloads.qi-hardware.com/people/werner/tmp/flickernoise-20111212a.fbi15:40
barbu-uucpmailed you mine15:43
wpwrakhmm. yours dislikes "Making time"15:43
barbu-uucpand yours works for me ...15:44
barbu-uucpmh15:44
barbu-uucpwell, at least it's not some weird hardware bug ;)15:44
wpwrakit could be quantum-entangled with the binary :)15:45
wpwrakcould be memory corruption somewhere15:47
wpwrakwith your binary, "Making time" compiles in the patch editor. but not when compiling all the patches for "simple mode"15:48
wpwrakvalgrind assures me that the parser is not to blame :)15:49
barbu-uucpand not a single problem with yours?16:00
wpwrakIce Drops looks a little odd. but maybe that's just my memory playing tticks on me16:08
barbu-uucpmwalle: btw, what's the reason for not including CFG_OUT/CFG_IN into urjtag?17:39
kristianpaulsince demo firmware delete, so all trust to rtems now? :)19:27
wpwrakwe trust flickernoise. and our ability to win against the bugs of rtems :)19:30
kristianpaul:)19:42
wpwrakso of all the libs milkymist builds, FN only uses libfpvm19:46
wpwraknow, let's see what the BIOS uses19:46
wpwrakhmm, hpdmc, base, hal, net19:48
wpwrakbarbu-uucp: plan for today: also add compiler warnings to milkymist. maybe this alone will be enough to reveal the source of the problem19:55
wpwrakbarbu-uucp: btw, what does the nick mean ? have you found some retro-computing equipment ? :)19:55
kristianpaulhe is using a uucp to tcp/ip link to gerthere :)20:03
kristianpauls/ger/get20:08
barbu-uucpwpwrak: btw, we can probably get rid of libmath20:58
barbu-uucpand of the soft-fp too20:58
barbu-uucpthe only place where it's needed now is the sinf() for the test program20:59
barbu-uucpwe can replace it with either a sinus LUT - to take care of Adam's ears during testing - or a more rash square/sawtooth wave21:00
barbu-uucpor a fixed-point sinus, but I'm not familiar with those21:04
wpwrakLUT sounds easiest :)21:05
barbu-uucpor just copy sinf() from libmath into the test program :)21:08
barbu-uucpand keep soft-fp, which doesn't cause much trouble21:08
wpwrakan array and a few lines of perl still sound easier :)21:10
wpwrakplus, we can then even use more complex sounds. if adam some day decides that a sine wave is boring and wants to hear some techno track, it will be easy to accommodate that wish :)21:11
barbu-uucpnah, perl is more work21:15
barbu-uucpor http://www.youtube.com/watch?v=tCRPUv8V22o :)21:16
wpwrak(more work) you mean something like this ? perl -e 'print join ", ", map { int(sin($_/50)*127+127) } 0..314;'21:18
wpwrakcute ;-)21:20
wpwrakokay. one entire techno track in ~30 characters. i admit defeat.21:22
wpwrak(around 02:10)21:22
wpwraki'm just waiting for one that has vocals, too21:22
barbu-uucpthe 5:14 is pretty cool too21:24
mwalleorignally, (1) it wasnt included in the bsdl file (2) you cant specify a valid length for the DR register21:31
mwallebarbu-uucp: btw whats barbu ?21:31
barbu-uucpbearded :)21:32
barbu-uucpin french21:32
mwalleok21:32
barbu-uucpbut what happens with the length of the DR register when you add the instructions manually?21:33
kristianpaulget rid of libnet when having a 2Mpbs uart core too :)21:35
mwallebarbu-uucp: its always overwritten, feel free to submit a patch to the urjtag ml, i'll ack it then ;)21:35
mwallekristianpaul: well having a 4mbit uart, i get only about 50kbps with gdbstub21:35
mwallehavent tried it with serialboot yet21:36
kristianpaulhmm21:36
wpwrak6:50 is pretty neat for the size21:50
GitHub189[migen] sbourdeauducq pushed 3 new commits to master: https://github.com/milkymist/migen/compare/eee6980...0ea7a9b23:30
GitHub189[migen/master] fhdl: allow a namespace to be specified for Verilog conversion - Sebastien Bourdeauducq23:30
GitHub189[migen/master] wishbone: only send ack to the active master in arbiter - Sebastien Bourdeauducq23:30
GitHub189[migen/master] wishbone2csr: fix double-write bug - Sebastien Bourdeauducq23:30
--- Tue Dec 13 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!