<rqou>
er, i'm going to need a more detailed mechanical drawing than that :P
<rqou>
they do have a datasheet, right?
<cyrozap>
They have more drawings on the item listing
<cyrozap>
And here's an AliExpress pro-tip: Even if they don't have the drawings (source code, datasheets, etc.) on the page, you can message the seller before they order--if they have it, they'll either send you a mini-CD with your item or send you a link after you buy it.
<cyrozap>
s/before they order/before you order/
rah has quit [Ping timeout: 255 seconds]
rah has joined ##openfpga
<rqou>
whitequark: I'm curious if you know what the debian texlive-lang-<foo> packages actually do?
<rqou>
(other than "be transitive dependencies of something else and take forever to install" :P )
<whitequark>
rqou: it literally says it in the description
<rqou>
"Support for (and documentation in) <language>"
<rqou>
why do I need that?
<rqou>
"This package includes the following CTAN packages: "
<rqou>
I also have no idea why I need that
<whitequark>
exactly
<whitequark>
it's an amalgamation of several CTAN packages
<whitequark>
well, why do you need a texlive-lang-X package?
<rqou>
it was a transitive dependency of something else?
<whitequark>
then look at that
<whitequark>
why do I have to tell this to you
<rqou>
I thought you might know off the top of your head since you seem to know how tex actually works?
<whitequark>
I don't see what's unclear here in the first place
<egg|zzz|egg>
wait, there are people who know how TeX works?
<egg|zzz|egg>
(knuth doesn't count)
<rqou>
huh, it isn't even a transitive dependency of anything apparently
<whitequark>
the only reverse depnedency of e.g. texlive-lang-cyrillic is texlive-full
<whitequark>
if you ask texlive-full to be installed, it will obviously drag in everything else.
<rqou>
I didn't. I guess apt was just being dumb
<whitequark>
there's the Install-Recommends setting.
<rqou>
I guess that must be it
<rqou>
great, debian just majorly broke something in my system
<rqou>
what do I need to downgrade today?
<egg|zzz|egg>
(if somebody knows how XeLaTeX works I'd love to understand why an empty delimiter that starts mathmode is scaled differently from an empty delimiter later in mathmode, whereas this isn't the case in LaTeX)
qu1j0t3 has quit [Ping timeout: 260 seconds]
<rqou>
I don't even know how xelatex is different from "normal" latex :P
<egg|zzz|egg>
Unicode \o/
<rqou>
great, debian is dumping me in "emergency mode" and I have no idea why
<egg|zzz|egg>
in particular you can use the unicode math characters from the SMP
<egg|zzz|egg>
e.g. "Let 𝜆 be a crossing-free left-alignment of 𝒮 on 𝑉."
<egg|zzz|egg>
(that's from my MSc. thesis, the version compiled with XeLaTeX)
<rqou>
AARGH WHY IS DEBIAN USELESSLY DROPPING ME IN EMERGENCY MODE?!
<rqou>
OK apparently the initrd is somehow hosed
<rqou>
whee bisection time!
<rqou>
followed by filing bugs that are probably going to get ignored
<rqou>
also why does latex need to do anything special with Unicode?
<rqou>
can't it just dump the input codepoints/bytes straight to the output?
<egg|zzz|egg>
wasn't it whitequark who tweeted something about TeX's oooold swich from 7-bit?
<rqou>
or is that not how PDF works?
<egg|zzz|egg>
that's not how LaTeX works though
<whitequark>
rqou: wat.
<whitequark>
tex is a layout engine.
<whitequark>
it has to handle fonts. (in fact it has its own font format, METAFONT, among others)
<rqou>
wait it does?
<whitequark>
sure it does
<egg|zzz|egg>
also XeLaTeX has (or has packages that have) support for modern font thingies
<egg|zzz|egg>
fontspec
<rqou>
(my experience with latex was limited to "hack this template until I can turn in my problem sets")
<rqou>
occasionally latex-compatible math markup would be used in other tools as well
<egg|egg>
I keep being weirded out at Unity not having a type for a (position, velocity) pair, nor a type for instantaneous rigid motion; is this common in phyiscs engines? or is it just a special case of me being weirded out by lack of strong typing?
<pie_>
probably the latter
<pie_>
its just a tuple!! :P
<pie_>
err 4-ple
<whitequark>
um
<whitequark>
that's still a tuple
<pie_>
2-ple
<whitequark>
"tu" in "tuple" is not for "2"
<pie_>
purple?
<egg|egg>
in french couple means ordered pair, and paire means unordered pair \o/
<whitequark>
egg|egg: why do you even have separate words for ordered and unordered pairs
<whitequark>
are you STL
<egg|egg>
whitequark: because Bourbaki?
<whitequark>
oh
<pie_>
heh
<pie_>
its a couple of paires!
<egg|egg>
pie_: it would be either a 6 or a 2 tuple anyway (6 dof, 2*3)
<egg|egg>
no, it's a couple of triples :-p
<egg|egg>
whitequark: which was a Principia version btw :D
<pie_>
well i was assuming a point for the forer, though for a rigid body ok
<pie_>
i cant find the "pay for everything you havent paid for yet" button on aliexpress
<egg|egg>
whitequark: Bourbaki also gave us the words injective, bijective, surjective; probably an improvement other the weird terminology that existed before (old german books have eineindeutig D:)
<egg|egg>
eindeutig means unique; eineindeutig would be... ununique?
<pie_>
more like uniunique?
<egg|egg>
maybe :D
<pie_>
idk what deutig means but ein is one
<pie_>
so im guessing thats supposed to be injective
<qu1j0t3>
latin 'copula' based on 'co-' meaning together; see also co-ordinate
LeelooMinai has quit [Client Quit]
<egg|egg>
cocycle!
<egg|egg>
I feel bad that this is what I had off the top of my head for an example of co-
<balrog>
feels like there's a hell of a lot of offtopic here these days
* egg|egg
hides
<whitequark>
egg|egg: I want to do one of those lesbian couple jokes but I don't know enough french to do it correctly >_>
cosmobird has quit [Ping timeout: 260 seconds]
egg|egg is now known as egg|tea|egg
LeelooMinai has joined ##openfpga
LeelooMinai has quit [Client Quit]
cosmobird has joined ##openfpga
<egg|tea|egg>
whitequark: hmm, what would that be like?
egg|tea|egg is now known as egg|egg
promach has quit [Quit: Leaving]
<whitequark>
egg|egg: there's the trope of someone asking a lesbian couple something to the extent of "which of you is the husband and which is the wife", which spawned a number of increasingly abstract parodies
<egg|egg>
ah, I had seen instances thereof, but I was unaware of the origin
<whitequark>
"which one of you is the lower bound and which is the upper bound" would be appropriate for an ordered set but I don't know enough french to translate that
digshadow has quit [Quit: Leaving.]
<egg|egg>
well not really lower and upper bound, it's not a range or an interval or somesuch, it's just that it has a first and a last element... hmm
<egg|egg>
aaah but maybe we can do something with a Kuratowski couple
<cr1901_modern>
`define NO_INFER_GP_SHREG_AT_ALL... I must've been creative with the defines that day
egg|egg is now known as egg|nomz|egg
kuldeep_ has quit [Read error: Connection reset by peer]
kuldeep_ has joined ##openfpga
<azonenberg>
cr1901_modern: lol
egg|nomz|egg is now known as egg|egg
digshadow has joined ##openfpga
<nats`>
egg|egg is french ? oO
<egg|egg>
and lives in zurich
<cr1901_modern>
azonenberg: btw, could you do me a favor and test that -DNO_INFER_GP_SHREG_AT_ALL still fails in your copy of yosys (currently am unable to test)?
<azonenberg>
cr1901_modern: Busy right now, poke me later in the day
<cr1901_modern>
ack
<cr1901_modern>
(It should fail b/c yosys will infer one GP_SHREG anyway despite me instantiating two; and only two GP_SHREGs exist)
<azonenberg>
Hmm yeah, i need to figure out how to cap shreg infrence
<azonenberg>
There should be some way to say, once you've inferred X shregs stop inferring them
<azonenberg>
or alternatively, do a 2-pass technique
<azonenberg>
first, find all the shregs
<azonenberg>
sort by size
<azonenberg>
then infer gp_shreg cells for the largest 2
<azonenberg>
(if you do it greedy you might extract two 2-cell shregs and not an 8-cell later on in the netlist)
<cr1901_modern>
azonenberg: Not saying this is a good usage of a gp4, but it can *physically* support up to a 16-bit shift reg. That's what my examples are testing
<azonenberg>
Yeah
<azonenberg>
What i mean is
<azonenberg>
shreg extraction should prioritize large shregs over small ones in the event there's not enoguh gp_shreg cells
<azonenberg>
And should leave smaller ones as discrete DFFs
<azonenberg>
then if you run out of DFFs, sorry the design doesnt fit
<azonenberg>
At some point down the road i want to add more aggressive techmapping
<cr1901_modern>
Lol... indeed
<azonenberg>
i think you can use some of the hard IP as flipflops and logic
<azonenberg>
i have some dirty ideas on how to abuse unused GP_COUNTx cells into logic functions
<cr1901_modern>
the counters can be used as FSMs, correct?
<cr1901_modern>
(or the "state" register of an FSM)
<azonenberg>
Kinda-sorta
<azonenberg>
they're up-down counters
<azonenberg>
you cannot jump to arbitrary states
<azonenberg>
you can reset to one specific state of your choice
<azonenberg>
and you can advance or backtrack, or remain in the current state
<azonenberg>
And you have to use the DCMPs to check which state you're in
<cr1901_modern>
That is actually probably fine for my purposes, maybe? Not sure
<azonenberg>
As of now i've had better luck doing FSMs with fabric logic
<azonenberg>
Over time i plan to add more support for inferring complex counter structures, DCMPs, etc
<azonenberg>
i have a PCB i want to get taped out in the next few days and might work on it after that
<cr1901_modern>
Short version, I wonder if a minimal uart (even half duplex) can fit into a GP4
<azonenberg>
Ooh
<azonenberg>
that would be interesting
<azonenberg>
And i still wanna try and fit more 10baseT into one
<cr1901_modern>
You have been warned: the retro part of my plans follow:
<cr1901_modern>
I still want to do a retro 8-bit SBC, and GP4 would be good for glue logic and a UART
<qu1j0t3>
ha. nice.
<cr1901_modern>
b/c well 16550 kinda sucks to program
* qu1j0t3
also wants to build some sbc's
<cr1901_modern>
One GP4 would do address decoding, interrupt source "wired-OR", and the other would be the UART. And maybe a third for a small bootstrap program so I could avoid having to put ROM on the board at all.
* cr1901_modern
shrugs
<cr1901_modern>
^ The retro part is above. You can uncover your ears now, azonenberg
<azonenberg>
Lol
<azonenberg>
hey, i was implementing 10baseT in a gp4 a while ago
<azonenberg>
that probably qualifies as retro :p
<whitequark>
that qualifies as a perversion
<whitequark>
:p
<azonenberg>
lol
<cr1901_modern>
How many GP4s does it take to screw in a lightbu- err, to implement 10baseT MAC?
<whitequark>
one
<whitequark>
well it's not really a MAC
cosmobird has quit [Remote host closed the connection]
<azonenberg>
Yeah i was barely able to do a PHY
<azonenberg>
and not a full PHY either
<azonenberg>
a full PHY would probably need two
<azonenberg>
one for the autonegotiation and bringup side, and one for the MII and manchester side
seu has joined ##openfpga
Hootch has quit [Quit: Leaving]
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
kuldeep_ has quit [Quit: Leaving]
kuldeep has joined ##openfpga
m_w has joined ##openfpga
<azonenberg>
whitequark: one day if i ever have the time
<azonenberg>
i want to try to figure out some way to automatically partition logic across many greenpaks
<azonenberg>
(like the big asic emulators do)
<azonenberg>
and make like a 6502 or something in a bunch of greenpaks
<azonenberg>
lol
kuldeep has quit [Quit: Leaving]
kuldeep has joined ##openfpga
<lain>
hah
kuldeep has quit [Remote host closed the connection]
<nats`>
someone will dump assets.nexperia.com/documents/ :D ?