Friday, 2021-12-03

programmerjakewell, before i was in college, when I still had spare time, I did dedicate several years to writing voxels, a game kinda like minecraft...i'm currently planning on working on it more at some point...i'm on rewrite #801:04
mikolajwI'm now completely unemployed and I have a lot of time. Does anyone have any easy starter task for me? That one with nmigen to C conversion had to be reassigned to someone else15:26
lkclmikolajw, no totally go for it, ghostmansd[m] is working on binutils18:14
ghostmansd[m]Yep, ack :-)18:15
ghostmansd[m]lkcl, cannot be closer to truth, looking at gas right now, lol18:16
lkcli do not do gambling, but in, when you see the results of all of the "if type == "VHDL" elif type == "BINUTILS" do this" statements18:18
lkclyou are going to freak out and go, "whyyy god, whyyyy did i quadruple the size of a 60-line function..."18:18
ghostmansd[m]Actually this enum is crap, I think there must be two classes18:18
ghostmansd[m]With a simple override18:18
lkclehn? que? enum? que?18:18
ghostmansd[m]This Format class18:19
ghostmansd[m]It's derived from Enum18:19
ghostmansd[m]Which I think is not really cool, actually this format check should be done only once18:19
ghostmansd[m]OTOH, this 60 line function dealt with one format, not two18:20
lkclohh yes, Enum class. very good at creating auto-numbering18:20
ghostmansd[m]So it's kinda expected there will be more code18:20
ghostmansd[m]But I agree not quadruple perhaps18:20
ghostmansd[m]The code sucks, yep, I know18:20
lkclthe thing is, pretty much the entire output changes.  header, footer, inside the for-loop, everything18:20
lkclif you used e.g. python-jinja (please don't!) there would be nothing common - at all - between the two18:21
lkclmikolajw, i'm provisionally allocating EUR 1200 to the task for you
ghostmansd[m]Hm. Never used Jinja, or even heard of it?18:23
lkclit's basically "let's provide everything that's awful about PHP... in python"18:24
ghostmansd[m]Saw it18:24
lkclyou can therefore embed python functions... in ASCII text.18:24
ghostmansd[m]How to unsee it?18:24
lkclthe syntax is {% .... %} and {{ ..... }} instead of <?php ?>18:24
octaviusDo a year of pure x86 assembly ;)18:25
lkclelectric shock therapy?18:25
ghostmansd[m]both together18:26
lkclwhat's that film where two actors brew up some memory-erasure concoction?18:26
ghostmansd[m]Don't even know which is worse18:26
lkclthey go, "i wonder if it's ready?" and faint18:26
lkcl1 hour later they wake up: "i wonder if it's ready? *sniff*" and faint...18:27
lkclrepeat about 3 times :)18:27
ghostmansd[m]Would be useful considering the profession18:27
lkclit was either Pineapple Express, or it was Green Hornet, or something equally stupid18:27
mikolajw  lkcl: but you said Dmitry takes it18:43
lkclmikolajw, no, we asked if it was *possible* for Dmitry to take it18:49
lkclhe chose instead to do binutils18:50
lkclso it's yours18:50
programmerjakelkcl I fixed the budget arithmetic for #241, please double check18:51
lkclah great.  i subtracted 1200 from 3700 and must have thought i'd subtraced 1500. thx for checking18:52
mikolajwlkcl: great, thanks, getting to this now18:53
* lkcl just trying to recall the os.environ variable in that can be used to output the actual python code-fragments18:54
lkclah ha! found it18:55
lkclyou can see, like 126-200 or so, how brain-dead-simple the decision-making is.  "if value.operator == "+" output LHS + RHS"18:59
lkcli would strongly suggest using functions the entire time (just like line 150 and 152, zdiv, zmod) - an add(LHS, RHS) function, a sub(LHS, RHS) function etc. etc.19:00
lkclbut, for goodness sake, *start small*.  start with *really* small nmigen HDL programs.19:00
programmerjakeblender on libre-soc?
programmerjakealready closed...19:15
tplatennoticed the changes 71 sec ago in set align_intr from request which comes...19:21
tplatenI have been thinking of running Godot4 on the libre-soc. Currently it does not even compile on my Talos II.19:23
lkcltook them 3 minutes to close it, deep joy19:24
lkcltplaten, yes.  that made an actual exception work, which is pretty stunning19:25
lkcl python3 simple/test/ nosvp64 >& /tmp/f119:25
lkcli need to confirm that19:25
lkclexc_happened is actually getting through to the PowerDecoder219:26
lkclcesar, ^19:26
lkclMSR is getting set to 0x8000000000001 which looks plausible.19:28
lkclHA!  PC is set to 0x600!19:28
lkclthat's fantastic news19:28
lkcltplaten, wha-hey!  case_5_ldst_exception works!  will just commit that now.  PC gets set to 0x300 due to "invalid"19:33
tplatenI have written a replay to your kanban mail from today.19:34
lkclsaw it19:36
lkcltplaten, yeah, hardware is tricky.  anything in m.d.comb creates a chain19:36
lkclgate->gate->gate->gate ...19:37
lkcland at some point, that chain is too long (too many transistors triggering other transistors that trigger other transistors...)19:37
lkclso you "capture" the output into a Flip-Flop (aka D-Latch, aka DFF, aka "register")19:38
tplatenI think I now understand, but it took me some time to understand. Too much time I think19:39
lkclthere's quite a lot going on, working indirectly here.19:40
* lkcl needs tea19:41
tplatenI now have a look at === alignment error (st) === as this was the first place where I found out that something goes wrong.19:44
*** tplaten <tplaten!> has left #libre-soc20:39

Generated by 2.17.1 by Marius Gedminas - find it at!