califax has quit [Remote host closed the connection]
califax has joined #openscad
<JordanBrown>
teepee, guso78k: since measurement is a UI feature only, with no language implications, it doesn't have to do everything or even have a clear path to doing everything on its initial integration. It's clearly a case where half a loaf is better than none.
mmu_man has quit [Ping timeout: 252 seconds]
erectus has quit [Remote host closed the connection]
erectus has joined #openscad
epony has quit [Remote host closed the connection]
epony has joined #openscad
LordOfBikes has quit [Ping timeout: 255 seconds]
<lf94>
teepee: roger
LordOfBikes has joined #openscad
erectus has quit [Remote host closed the connection]
erectus has joined #openscad
J23k5 has quit [Quit: Client closed]
J23k5 has joined #openscad
J23k5 has quit [Client Quit]
epony has quit [Remote host closed the connection]
epony has joined #openscad
snaked has joined #openscad
TheAssassin has quit [Remote host closed the connection]
<kintel>
pca006132 In term of openscad interpreter: Not sure if this is helpful but I tinkered a bit with the idea of splitting off the parsing + language eval parts into a separate lib a long time ago. The sticking points were use/include semantics, but if we clean those up (at a small cost of dropping corner case compatibility) it can become pretty clean I think
<kintel>
A long time ago, I wrote an OpenSCAD parser+evaluator in javascript. It used to pass all non-graphic OpenSCAD tests: https://bitbucket.org/openscad/openscad-js
<kintel>
(back when javascript was still a good idea, anything compiling to wasm would probably win today)
<teepee>
same as earlier link? then that's great, already pushed to repo. just needs some colors, I can do that
<J23k>
lf94 if you make the cannon completely in 2D you can add some ±offset to smooth and then rotate_extrude
<JordanBrown>
teepee: OK, no problem. I debated that , but I've recently felt that spaces make the names easier to read and the only way to get people to always handle them is to use them.
<JordanBrown>
Scopeuk: I'm not quite clear on whether you think that spaces-in-paths should be totally routine by now (and so not something we have to think about), or should just never be done (and so not something we have to think about).
<JordanBrown>
teepee: I forgot that you might have scripts to manage this stuff, that might need to take care.
<teepee>
I generally agree, but I don't want to risk door 24 breaking ;-)
<JordanBrown>
yes
<JordanBrown>
Of course, if you want to bring in a whole additional set of headaches, just have file names that start with "-".
<InPhase>
lol
<teepee>
I don't think that would be an issue but I'm still not going to try :P
<InPhase>
Ah good, we are filled up to 24.
<JordanBrown>
Other than forcing a name to the top of an alphabetical list, there's not really any readability gain from starting a name with "-".
<teepee>
also I really trained to quote filenames in scripts so there's a good chance spaces would be just fine
<JordanBrown>
Yeah, I try to do that as a matter of routine too.
<JordanBrown>
And for super extra fun, there's names with newlines in them.
<teepee>
no, but it's fun passing it to linux commands for people who don't know what's going on ;-)
<InPhase>
As a Linux user I object to spaces in filenames, even though I write my scripts carefully to handle them.
<InPhase>
They make quick scripts done on the fly more cumbersome, which is the primary objection.
<JordanBrown>
There's no usability rationale for newlines (or other control characters) in paths, but UNIX-derived OSes allow it and so any general-purpose tool has to handle them.
epony has quit [Remote host closed the connection]
<JordanBrown>
Really the kernel should reject them. Near zero compatibility impact, significant security improvement.
<JordanBrown>
But until that happens, find -print0 and xargs -0.
<InPhase>
JordanBrown: Although ironically, the last issue I had with spaces in filenames was this week while remote supporting a Windows user in Texas. We were trying to debug a network connection issue using netcat, and the user was reporting that they couldn't run ncat on the command prompt, even though the file browser showed ncat right there in the directory!
epony has joined #openscad
<InPhase>
JordanBrown: We got lucky on figuring it out, because I had them run "dir" at the command prompt rather than use the file browser, and learned that "ncat .exe" was indeed in that directory.
<InPhase>
JordanBrown: And because Windows by default hides the file extensions, the browser showed file "ncat " with an invisible space...
<JordanBrown>
Ouch.
Guest52 has joined #openscad
<JordanBrown>
Yeah, it would be a win if it visually highlighted leading, trailing, and doubled spaces.
<JordanBrown>
My IRC client is using a proportionally-spaced font, and so "ncat .exe" has a very thin space. I wasn't absolutely sure that there was a space there until I ran the mouse cursor across it, selecting.
<Guest52>
Respected Sir/Madam,
<Guest52>
I am Utkarsh Upadhyay, a student pursuing Information Technology at KIET Ghaziabad. I am in my second year of studies. I have some experience in open source contributions and I'm well-versed in Git. Additionally, I have practical knowledge of Python and have completed basic machine learning projects. I'm also familiar with AWS and Docker. I am
<Guest52>
eager to make a positive contribution to your organization.
<Guest52>
Could you please guide me on how to get started with GSOC and become a part of your esteemed organization? I am excited to learn and contribute to meaningful projects under your mentorship.
<Guest52>
Hoping to hear from you soon.
<Guest52>
Regards,
<Guest52>
Utkarsh Upadhyay
<InPhase>
Guest52: Is that addressed to us?
<InPhase>
Guest52: OpenSCAD is primarily a C++-based project.
<J23k>
nice .. these helical things can suck for printing
Guest52 has quit [Quit: Client closed]
<teepee>
tried on the side which worked except one of the arms coming loose
<J23k>
had some that could be printed only mirrored as the direction made a difference (and can't be altered in slicers)
<teepee>
and the support worked quite well
<teepee>
oh, interesting
<teepee>
trying now upside down streched out a bit more
<J23k>
oh yes with supports is easier so not much deflection
<teepee>
basically cutting out the hook in the slicer
<JordanBrown>
I have occasionally wished that the slicer would let me give very specific instructions about how to handle a particular piece.
<J23k>
also using 2-3mm z-hop is helpfull as the thermal stress in these helical arms lead to height differences and your hotend colliding
<teepee>
yeah, the "place on face" in prusaslicer is a good start, but at most 5% of what one could wish for :)
<JordanBrown>
Like where I have a 10x100mm gap that needs to be bridged, and it's choosing to bridge it length-wise.
<teepee>
yeah, I increased from 0.2 to 0.4 for now
<JordanBrown>
There's the global bridging angle, but that doesn't help if you have two such gaps.
<JordanBrown>
at right angles.
<teepee>
oh, yes, bridge support seems to be quite bad still, I've not seen anything useful in cura, prusaslicer might be a tiny bit better but still far from clear in detecting bridges
<J23k>
normally you can increase z jerk and speed up to 30 .. leading to much faster printing and better quality
<teepee>
I'm running the input shaper config, I think that has Z at 30
<JordanBrown>
Prusaslicer *detects* them OK, in that it draws them in blue. But then it sometimes decides to handle them stupidly.
<teepee>
ah, no, 12 actually
<J23k>
ah ok .. since i adjusted mine it is just "clack" "clack" and you can't see the spindle moving - it is like jumping into the new position
kintel has joined #openscad
<teepee>
good point, yes, I think I mostly saw them in blue but no useful handling
<JordanBrown>
Yes. And I think I've seen cases (maybe a reduction in size in a vertical cylinder hole) where it decided to draw an inside perimeter before drawing the bridges.
<kintel>
peeps ..but after my first round I think I need to upgrade from bash to python. I just find shell so much nicer for calling into external tools...
<teepee>
reasonable start https://imgur.com/a/ybV9ePq but it's thin and quite high, so lots of potential for messing it up
<InPhase>
lf94: I fixed up the preview on your cannon. You had 0.0005 as your overlap, which doesn't always parse as adequate in the different preview engines at all zooms and angles. I made it height+1 for 0.5 overlap (and committed that change), because there's no loss giving it plenty of extra, and now the wheels look perfect.
<InPhase>
lf94: It seems to be the case that these sorts of differences vary by different drivers. Preview is an inconsistent beast.
<JordanBrown>
Yeah, I start out using something like "epsilon=0.01" and adding epsilon to sizes, and eventually realized that +1 was a lot terser, was not a problem for the geometry, and that by convention I would never be adding a constant 1 to something for any *other* reason. Similarly +2.
<JordanBrown>
teepee what is that pointy dark gray thing on the left?
<teepee>
fan shroud
<JordanBrown>
OK, thanks.
<InPhase>
Yeah. I do +1 for all invisible negative pieces, and translate -1 height +2 for two-sided negative pieces that aren't centered. I go with 0.01 or 0.001 only for pieces that would create a functional difference in sizing, or that risk colliding with another nearby element.
<JordanBrown>
I do use +0.1 or +0.01 when I'm just "painting" a color onto an existing object.
<JordanBrown>
Like the trim paint on the teacups.
<JordanBrown>
Another trick is *3, center=true. That punches through the object no matter which way everything is aligned.
<InPhase>
Yeah, same. And I noticed that while inspecting the teacups. :)
<InPhase>
I also like the *3 one. :)
<JordanBrown>
I am always interested in code review comments.
<InPhase>
I'd be happy to provide them! But I inspected only the output by zooming.
<lf94>
InPhase: nice to know - I knew adding an epsilon usually fixes that, but that seems to be quite a bit amount :)
<lf94>
J23k: feel free to pretty it up along with teepee ;p
<lf94>
It's CC0 so go crazy
<J23k>
I would end in making it completely new that would be easier as the cannon is the main part
<lf94>
gotchya :)
<teepee>
maybe a different one with some animation next year?
<InPhase>
JordanBrown: Glancing at the code, the only thing I think I might have done differently, is use a formula for well-spaced colors that I have instead of the large personShirtColors table. But... I am not sure where I put that formula. :) And I see that your approach targets the goal of omitting the teacup colors.
<JordanBrown>
InPhase you would think that after almost 40 years of C my C style would be pretty well nailed down, but I still tweak it slightly on occasion.
<InPhase>
Oh, actually you do duplicate some of the teacup colors.
<JordanBrown>
I wasn't sure what formula yielding RGB would yield colors that people might actually like
<J23k>
teepee the twist thing is missing an image
<JordanBrown>
.
<InPhase>
Yeah, pretty is tougher when you have so many colors going on in an image.
<teepee>
I know, update coming in a minute
<teepee>
I combined 2 images I created the 3MF files for
<JordanBrown>
Also I get the impression that "good color" is not evenly distributed throughout the RGB space.
<InPhase>
JordanBrown: It gets even harder when you're making scientific tools and then you run into all those colorblind scientists who need different choices.
<JordanBrown>
Yeah, for accessibility you pretty much need to avoid carrying any critical information in color.
<InPhase>
If you start spacing out around the hue space in order, the first two you pick are red and green, and then boom, already failed for the most common RG colorblindness.
<JordanBrown>
teepee: drat, never got around to pushing Nessie. The calendar is full, right?
<InPhase>
It has 24, and 90% of them are lamps. :)
<teepee>
2023 yes, 2024 has 24 open slots :)
<JordanBrown>
teepee: so I figured.
<JordanBrown>
And if we ever get #4478 merged, I have a train set for 2024.
<JordanBrown>
But it relies on being able to easily make lists of geometric objects.
<JordanBrown>
Nessie is relatively printable. You need to tip the head and tail maybe 30 degrees to avoid overhang. But the head is badly off-balance and falls over, e.g. if you just want to put it on your mantle.
<JordanBrown>
So the current design leaves a no-infill void at the bottom and stops at the top of that void so I can put in some lead weight.
<JordanBrown>
So far so good. But for some reason you get a lot more stringing than I do.
<teepee>
and I tuned down temp from 220 to 195 already
<JordanBrown>
I don't know exactly what I've done to reduce stringing, but these days I rarely get any.
<teepee>
from the prusament pla default to this DasFilament PLA
<JordanBrown>
My preferred filament recently has been Overture PLA, because (a) it's cheap (b) I can subscribe to it on Amazon and (c) it comes on a cardboard spool so I don't feel guilty.
<JordanBrown>
It would be nice if a program could recommend FPS and number of frames.
<JordanBrown>
FPS not so much, since that's mostly a quality/performance/size tradeoff.
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<JordanBrown>
Actually, let me back off, not frames, seconds.
<JordanBrown>
Most of my animations are designed for a particular number of seconds, and it's a nuisance to have to coordinate the SCAD program with the script that runs the openscad to generate the frames.
<JordanBrown>
That will get more important if OpenSCAD ever gets .gif and .webp generation as built-in features.
<t4nk_fn>
hehe... teepee .. are you trying to print a spiderweb? :b
<JordanBrown>
I suppose I could have a general animation driver script that first runs the program and asks it for the parameters that it would like, then uses those parameters.
<teepee>
pah, spiderweb is probably much easier to print than that thing
<t4nk_fn>
hehe
<t4nk_fn>
is anyone using the gentoo live build for openscad here btw? I think I ran into some error building last night
<teepee>
I don't, but if you pastebin the error we can have a look
<t4nk_fn>
I think the only hint is: fatal: update_ref failed for ref 'refs/git-r3/media-gfx/openscad/0/submodules_manifold/bindings_python_third_party_nanobind/__mai
<JordanBrown>
But it looks like 2024 is open for business!
<teepee>
yep, plenty of place for 2024
<teepee>
who knows, maybe we can fill it before december 1st next year :)
<teepee>
t4nk_fn: pastebin the full log? maybe there's some earlier hint
<t4nk_fn>
teepee, yes, I thought of that too (was off cloning that nanobind) ... the log is riddled with color codes orsth, let me see if I can clean it up before pasting
<JordanBrown>
teepee that matte brown has a nice wood look to it, at least in the photo.
<JordanBrown>
2024: First!
<JordanBrown>
Not that it's really up to my standards for release, but it gets the ball rolling.
<JordanBrown>
I have time to review and clean up.
<JordanBrown>
Animating the waves and moving Nessie across them might look good.
<JordanBrown>
Hopefully by then I will make surface() take an array as input, so I won't need the sfc() module that is a large fraction of the total.
<t4nk_fn>
argh, teepee, couldn't get rid of it that fast, it's not too long though: https://dpaste.com/5B2SMQ9K3
<t4nk_fn>
yeah, I cloned that nanobind standalone, and it told me about the --recursive indeed
<teepee>
so that looks like a submodule of a submodule of a submodule :)
<t4nk_fn>
yeah hehe
<teepee>
like openscad->manifold->nanobind->robin_map
<JordanBrown>
I hadn't realized that anybody was using the Matrix room after the bridge washed out. I'll start watching it.
<teepee>
never got feedback from the matrix ticket
<teepee>
that's now 2 month old or so
<teepee>
:(
<JordanBrown>
Matrix and Libera are very aware of the problem.
<JordanBrown>
Something went bad in the bridge, they tried to fix it, failed.
<JordanBrown>
I don't know whether anybody is still trying.
<JordanBrown>
But something happened on the Libera side that the Matrix folks didn't like, that was causing problems, and when the Libera folks couldn't get it fixed the Matrix folks turned it off.
guso78k has joined #openscad
<JordanBrown>
Oops, got that backwards. It was the Libera folks who pulled the plug because of problems on the Matrix side.
* t4nk_fn
quickly revisits the libera-ops homes he just egged and picks them up and cleans the windows
<t4nk_fn>
this is how wars start ;)
<t4nk_fn>
oh no, wait
* t4nk_fn
eggs them again!
<JordanBrown>
Best you read up on the history and make your own egging decision.
<teepee>
the concerns are probably quite real, the only annoying thing is there's no simple way to recover after that deadline happened
<teepee>
oops, just saw that new print is estimated 8h
<JordanBrown>
teepee my impression is that there are two levels of bridging, and that they started with the most "expansive" level, and when there were problems they backed off to the lower level, and there were still problems with that so they shut it down entirely.
aks12o4 has joined #openscad
<aks12o4>
Hi everyone My name is Ashish Kumar Singh. I am from Delhi, India. I am new to open source and am excited to learn and contribute to the open source community. I am familiar with C, C++, Python and have a little bit of know how of JavaScript. I have previously worked on Machine Learning projects. I am looking forward to connect with all my fellow
<aks12o4>
developers and would like to learn and grow with them alongside. Any leads on how to get started in this space are highly welcomed.
<teepee>
aks12o4: simple first step, follow discussions here, see if there's something that sounds interesting
<aks12o4>
teepee Ok :]
<JordanBrown>
aks12o4: do you use OpenSCAD? If not, it's likely to be unrewarding - you working on a project that doesn't really interest you, and others having to educate you on the project so that you can work on it reasonably.
<aks12o4>
JordanBrown I have previously never worked on OpenSCAD.
<JordanBrown>
It seems odd to work on an open source project that you do not yourself use. I suppose that experience is experience, but if you don't use it yourself then experience is *all* you're getting... one would think that you could find a project that matches your interests.
<JordanBrown>
And if you don't use it yourself, how will you understand its needs?
<JordanBrown>
I'm not trying to discourage more people from working on OpenSCAD, but it seems odd. It's not like open source is a job that you do even through you don't like it, just to get a paycheck.
<teepee>
no, it's fine, it's pretty much what I meant too. I assume it's about Google Summer of Code, where there is not a requirement as such to be too involved with some specific project
<teepee>
but in my view it's a huge plus for motivation and also results if there's some sort of interest in addition
<aks12o4>
JordanBrown How do I find a project that matches my interests then? :] Also thanks for your words. I am always open to feedback as I told you I am new to this and open source in general.
<JordanBrown>
Of course, the interest doesn't have to be in CAD per se. One might be interested in, say, 3D graphics.
<teepee>
yes, or math in general, openscad can work as a slightly strange graphics calculator too :)
<t4nk_fn>
aks12o4, ... you could Start a project of your own ;)
<teepee>
t4nk_fn: that's always an option in general, should it be about GSoC, then no
<JordanBrown>
What software do you use today - especially stuff you use for fun? Is it open source? If it's commercial, is there an open source alternative that you could switch to, or work on with the goal of making it a viable competitor to the commercial product?
<teepee>
well, at least not directly, as only "established" projects can take part at least to the rules in the past years
<aks12o4>
I am just exploring. I don't know if I have an interest in Computer Graphics yeah but have always been passionate about Mathematics.
cart_ has quit [Ping timeout: 252 seconds]
<aks12o4>
teepee If I am being honest its about GSoC
<teepee>
then now is actually the perfect time to explore that
<aks12o4>
how to get started for the same?
<teepee>
exactly what was said already. ideally find something which overlaps your interests just as JordanBrown suggested
<teepee>
if you have maybe a friend who owns a 3d printer, that could be OpenSCAD
<teepee>
if it's math, check out the list of projects that took part last year and see if something jumps out
<teepee>
at this point, nobody knows who will take part in next years GSoC, it's announced to happen, but project selection will be beginning next year, I assume (same as in past years)
<aks12o4>
where to find something which overlaps my interests say, also what are the next set of steps I should take after that.
<JordanBrown>
For me, I got involved with OpenSCAD development because I was using the program and finding that it didn't do things that I wanted, and could be made to do them.
<JordanBrown>
If it wasn't OpenSCAD, I'd probably work on Thunderbird.
Guest47 has joined #openscad
Guest47 has left #openscad [#openscad]
<aks12o4>
For as far as I am familiar about Open Source development I know it off like resolving open issues in the codebase of an organization on their GitHub repository.
<aks12o4>
I know I sound naive. I hope you understand.
<aks12o4>
What are the first set of steps this community expects off from someone who is just getting started and wants to get involved?
<JordanBrown>
Look at the github issues list, or use the program and come up with an issue of your own. If it's simple, propose a fix via a pull request. If it's not simple, discuss and then propose a fix.
<JordanBrown>
If it has to do with changes to the language, it's automatically not simple :-)
<aks12o4>
JordanBrown Ok thanks a lot :]
teepee_ has joined #openscad
teepee has quit [Ping timeout: 240 seconds]
teepee_ is now known as teepee
aks12o4 has quit [Quit: Client closed]
aks12o4 has joined #openscad
aks12o4 has quit [Client Quit]
<peeps>
kintel, minkowski seems unnaturally over-represented in ochafik's tests
<peeps>
which is i guess meant as a way to emphasize manifold's strengths? but if so, that still seems misguided it relies on CGAL's convex decomposition. why not just show it doing heavy CSG boolean ops that it is good at
<InPhase>
peeps: I seem to recall he struggled with minkowski, which could explain it having extra tests.
<guso78k>
pca006132 right now in install the python embeddable package's content into a directory called 'python_embed'next to openscad and only a link of python3.11.dll next to openscad, so openscad.exe finds the dll and it starts but it does not see the python std lib.
<peeps>
the description admits that "Note that minkowski operations get a specific boost thanks to the introduction of parallelism in the algorithm itself (and then the union of parts it generates benefits from Manifold's own parallelism)"
<peeps>
i don't know, i find it hard to follow a repository as a gist. and that maze.scad is obnoxiously long
<guso78k>
when copy all the zip file contents next to openscad.exe it does work, but nobody would like the python stuff in such a mess next to openscad and i was not able to tell python about the python_embed subfolder with quite some trials setting variables in PyConfig_SetByteString.
<peeps>
but like it has impressive enough performance benefit to regular CSG ops that I don't see the need to put minkowski at the forefront of every benchmark
<guso78k>
Today ,when googling for a solution to this issue, it became clear to me that even i get the embeddable package working: it wont't have a pip as-is and I see the ability to leverage from external packages as a key benefit from using python.
<guso78k>
and even when i manage to add a pip to it, you still have to add all your desired pip packages, so why dont' use the system python packages from the very beginning ?
<InPhase>
peeps: Oh, I see my threads made it into ochafik's benchmarks. :)
epony has quit [Remote host closed the connection]
<InPhase>
It's not obvious what he ran for that though. I guess maybe the built-in demo. If so, that's basically a bunch of high count manual polyhedrons with some differences and unions.
JordanBrown has quit [Ping timeout: 268 seconds]
epony has joined #openscad
JordanBrown has joined #openscad
JordanBrown has quit [Read error: Connection reset by peer]
JordanBrown has joined #openscad
JordanBrown has quit [Ping timeout: 260 seconds]
JordanBrown has joined #openscad
JordanBrown has quit [Read error: Connection reset by peer]