TheAssassin has quit [Quit: No Ping reply in 180 seconds.]
TheAssassin has joined #openscad
<peepsalot>
Scopeuk: updating the wiki page you mean? I'm hoping it will get fixed on msys side soon enough, that we can go back to their official package, but if you feel like editing it be my guest
<peepsalot>
Scopeuk: the macos thing being the folder permissions issue?
<Friithian>
peepsalot: after downloading users may need to right click on the application and click enter to be able to override macos' built in security
J1A846896478096 has joined #openscad
J1A8468964780 has quit [Ping timeout: 260 seconds]
pa has quit [Ping timeout: 255 seconds]
<InPhase>
Oh. Missed the primary school teacher. There are many other good tools to use there, like blockscad3d.
<dalias>
scopeuk, should i report that on the tracker?
pah has joined #openscad
califax has quit [Remote host closed the connection]
califax has joined #openscad
fling has quit [Remote host closed the connection]
fling has joined #openscad
califax has quit [Remote host closed the connection]
califax has joined #openscad
califax has quit [Remote host closed the connection]
<InPhase>
Yeah. We really shouldn't have that behavior.
<InPhase>
I haven't looked enough into the preview logic to know if this is unfixable or a resolvable quirk of how color is propagated. But I am pretty confident we could do it right with render-propagated color.
<peepsalot>
opencsg figures out the depth buffer of which surfaces are shown. then we go through and paint the primitives which exactly match that depth value
<peepsalot>
so, last paint color wins
<Friithian>
urgh, outer ring didn't fit over tube. have to resize and reprint
<JordanBrown[m]>
preview is black magic. intersection support in preview doubly so.
J1A84 has quit [Quit: Client closed]
J1A84 has joined #openscad
<JordanBrown[m]>
I understand how union support works, and kind of understand how difference support works. I don't understand at all how intersection support works.
<InPhase>
peepsalot: Ah. Well, that is pretty fundamentally sketchy I suppose.
<peepsalot>
it is explained some more in the issue #1000 which I marked it as duplicate of
<InPhase>
If opencsg really has no knowledge of color, then we would need something like tree-like evaluation of surfaces just to figure out which colors to preserve.
<InPhase>
Which sounds like a bit of overhead to get colors to work out right.
<peepsalot>
like theoretically the stencil buffer might be usable in some way to track surface "id" or something (and later match to a color), but I think OpenCSG uses that interally also.
pa has quit [Ping timeout: 252 seconds]
pah has joined #openscad
<Scopeuk>
yeh, once we got into it I guessed it would be a duplicate
<JordanBrown[m]>
Note that if you make the big red cylinder just a smidge taller its Z-fighting disappears... and that the part of the small red cylinder that's intersected away still shows up in Z-fighting.
castaway has joined #openscad
<JordanBrown[m]>
If you render() intersection() {...} then the big red cylinder disappears, but you still have the union-derived Z-fighting for the small cylinder, where it can't decide whether that part is green or red.
<JordanBrown[m]>
fixing *that* is much harder.
Junxter has joined #openscad
<JordanBrown[m]>
To get color-safe union you need to render a difference, and then union. And you need to color that rendered difference.
<Friithian>
you ever set $fn to be something way higher than it needs to be for no good reason?
<InPhase>
Friithian: I sometimes use that for stress testing.
<InPhase>
Or just to see pretty things, if you really want no good reason.
<Friithian>
this is just a cap for a pipe lmao
<InPhase>
For printing I typically just stick $fa=1; $fs=0.4; and move on. Or sometimes but rarely I'll upgrade to $fa=0.5; $fs=0.2;
<InPhase>
I use $fa=1; $fs=0.4; so often that I think these should be the program's defaults.
<JordanBrown[m]>
My winner for doing fine-grained work that ended up unnecessary was when I was laboriously constructing a polyhedron (one of the first times I ever did, which is why it was laborious) for the point where two bevels met, only to realize that that pyramid was going to be about one cubic mm.
<InPhase>
JordanBrown[m]: Do you ever look at the printed 1mm blob and reminisce about how it should look?
<JordanBrown[m]>
I realized that the detail work was unnecessary *after* I finished it. So the model has it right.
<JordanBrown[m]>
And it's a bevel underneath something, as a designed-in support... and so it's not visible.
<JordanBrown[m]>
*And* I realized that the better approach was to intersect two bevels to create the connection, rather than constructing a polyhedron.
<InPhase>
I've certainly wasted time designing sub-printable features, but I suspect your example was probably more of a time sink than most of my recognition failures at this.
<InPhase>
Sometimes you just get zoomed in and caught up in it, or you start your design at the wrong scale and figure you'll adjust the size later, and then you do and oops.
<JordanBrown[m]>
On a completely different topic, but sort of related to something that comes up in OpenSCAD every once in a while (precision of floating point numbers), I just got this quote and link from the This is True newsletter...
<JordanBrown[m]>
Other Good Reading: Pi (3.14159...) is used in a lot of mathematical computations. NASA uses it for navigation. Some people have memorized thousands of digits of its infinitely long sequence. But do we really need to know thousands and thousands of digits? NASA JPL’s Chief Engineer for Mission Operations and Science, Marc Rayman, answered that question in 2016, and recently updated it. His answer is understandable and interesting:
<JordanBrown[m]>
How Many Decimals of Pi Do We Really Need? from JPL.
<Friithian>
and I forgot to refill my pen again so I'm nearly out of ink :(
<JordanBrown[m]>
I don't feel like doing the math again to get the details, but 64-bit floating point is precise enough to measure the distance to from the Earth to the Sun to something like the nearest millimeter.
teepee has quit [Ping timeout: 255 seconds]
<Friithian>
if I wasn't trying to learn probability shite I'd do the math
aiyion1 has joined #openscad
Sauvin has quit [Read error: Connection reset by peer]
Bocaneri has joined #openscad
<Friithian>
oh ew I forgot this slicer doesn't randomize z seam
<JordanBrown[m]>
That sounds too close. Aw, now you're going to make me go do the math.
teepee has joined #openscad
<JordanBrown[m]>
ah, I slipped a prefix, and thus three orders of magnitude. 0.03mm sounds plausible.
<JordanBrown[m]>
Anyhow, plenty accurate for most practical purposes.
Guest186 is now known as Sauvin
milza has quit [Read error: Connection reset by peer]
milza has joined #openscad
milza has quit [Read error: Connection reset by peer]
<peepsalot>
i think i embedded a tiny glass splinter in my fingertip, and its actually below the surface so I don't see how I will ever remove it
<JordanBrown[m]>
Dig.
<peepsalot>
i wonder if it will eventually work its way out
<JordanBrown[m]>
On the other end of the spectrum, we have "half precision", 16-bit floating point with an 11-bit mantissa and a 5-bit exponent.
<JordanBrown[m]>
Measuring the distance from Los Angeles to New York to the nearest mile, or so.
<InPhase>
peepsalot: One strategy I've tried is dumping rubbing alcohol on it, which causes it to hurt enough that you don't notice the splinter for a while. I'm not sure this helps, I'm just saying I've tried it.
<peepsalot>
if you used fixed precision 64 bit number (say 1 unit = 1 micron), then you can represent a distance out to well past pluto
<JordanBrown[m]>
Fixed point is underrated.
<teepee>
clipper2 moved from fixed to floating :P
<peepsalot>
teepee: after looking at it a little more closely, it still converts to fixed. its just providing a slightly more convenient interface for double input
<teepee>
oh, so it's just a wrapper :(
J1A84 has quit [Ping timeout: 260 seconds]
<teepee>
but then, maybe better than the one we have
<JordanBrown[m]>
I have some tweezers sort of like that. The pointy tip is good for digging splinters out.
<peepsalot>
you have to provide a power of ten scale, which is worse granularity than the power of 2 i implemented
<J1A8438>
inorganic splinter are encapsulated .. wood will be moved out
<JordanBrown[m]>
Time to go back to pretending to work...
<teepee>
JordanBrown[m]: yep, they specifically sell it for that, and it did work quite well a couple of days ago with a very tiny tip of a thorn I noticed via an annoying red spot on the finger :)
Non-ICE has quit [Quit: Screw you guys! I'm going home!]
Junxter has quit [Ping timeout: 260 seconds]
J1A843857 has joined #openscad
J1A84385718 has joined #openscad
J1A8438 has quit [Ping timeout: 260 seconds]
J1A8438571814 has joined #openscad
J1A843857 has quit [Ping timeout: 260 seconds]
J1A84385718 has quit [Ping timeout: 260 seconds]
ur5us has joined #openscad
linext has quit [Read error: Connection reset by peer]
teepee_ has joined #openscad
teepee has quit [Ping timeout: 255 seconds]
teepee_ is now known as teepee
pah is now known as pa
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
ali1234 has joined #openscad
ali12341 has quit [Remote host closed the connection]
<peepsalot>
teepee: hrm the CI server actually contains cmake 3.24. So I think the only potential problem might be OBS for 18.04? if it is an issue could we use a PPA for cmake?
<peepsalot>
i was just going off of the version listed at packages.ubuntu.com
<teepee>
I'm not sure ppa would work, but there should be a way to get a newer cmake in somehow
<teepee>
it might be as easy as pasting 3 urls into a custom cmake package