teepee changed the topic of #openscad to: OpenSCAD - The Programmers Solid 3D CAD Modeller | This channel is logged! | https://openscad.org/advent-calendar-2021/ | Website: http://www.openscad.org/ | FAQ: https://goo.gl/pcT7y3 | Request features / report bugs: https://goo.gl/lj0JRI | Tutorial: https://bit.ly/37P6z0B | Books: https://bit.ly/3xlLcQq | FOSDEM 2020: https://bit.ly/35xZGy6 | Logs: https://bit.ly/32MfbH5
ur5us has quit [Ping timeout: 240 seconds]
\mSg has quit [Ping timeout: 260 seconds]
<InPhase> OlivierChafik[m]: So far the worst thing about testing fast-csg is having to do the renders with it off...
ferdna has quit [Quit: Leaving]
use-value has quit [Ping timeout: 250 seconds]
use-value has joined #openscad
J22 has quit [Ping timeout: 256 seconds]
drfff has joined #openscad
<OlivierChafik[m]> InPhase: hahaha! I keep waking up to benchmark runs where only the baseline is still running
ur5us has joined #openscad
<InPhase> OlivierChafik[m]: I'm testing my hair comb right now, which is probably my most poorly made design in terms of efficient use of OpenSCAD. https://www.thingiverse.com/thing:1689741/files
<othx> InPhase linked to "Ergonomic Hair Comb With Personalized Text by rcolyer" on thingiverse => 3 IRC mentions
<InPhase> (Although a very popular download.)
<InPhase> I was a newbie when I wrote it, and I really went to town with those minkowskis.
<InPhase> OlivierChafik[m]: I suppose my top actionable complaint is that right now fast-csg is super noisy. I'm getting on the scale of a thousand lines of output on the Console, which is prohibitively large. There are often times I want to compare echo outputs from one run to the next, and that's made impossible with fast-csg on.
<InPhase> Probably these outputs could be moved under the debug preference checkbox.
<InPhase> Hmm. Does trust-fast-csg always terminate?
<ccox> halting problem -- cannot be answered. ;-)
<InPhase> Fair. ;)
<InPhase> This design is about 13 minutes in 2021.01, 4.5 minutes with fast-csg, and I'm at over 20 minutes with trusted fast-csg.
<ccox> does it complete in bounded time... that depends on how it traverses the object tree.
<InPhase> I'm at 12GB and 100% cpu and chugging away. I guess I'll let it go until I run out of patience.
<InPhase> [fast-csg] corefinement mesh union (1297966 vs. 1017596 facets) This number seems to be steadily rising.
<OlivierChafik[m]> InPhase: I'll send a PR to reduce the normal noise *and* increase it when the debug feature is on
<OlivierChafik[m]> (thanks for the feedback)
<InPhase> *thumbsup*
<OlivierChafik[m]> (hah reading one of your messages at a time)
<OlivierChafik[m]> yeah the remeshing thing is really really gonna fix these cases up https://github.com/openscad/openscad/pull/4107/files
ferdna has joined #openscad
<OlivierChafik[m]> can't wait for it to land in CGAL
<ccox> of course, if your tree traversal creates more branches, we may be back to the Collatz Conjecture :-P
<OlivierChafik[m]> (that one model shared one twitter was getting exponentially larger parts it seems, impossible to complete; with the remeshing branch, it got 4x faster than the original)
<OlivierChafik[m]> InPhase: your model might never finish. For a fraction of the CPU power, maybe you can compile OpenSCAD with the remeshing support:
<InPhase> It is indeed very slowly growing in RAM. 15GB now.
<InPhase> About 50 minutes of runtime.
<InPhase> I'll give the remeshing a try.
<lf94> finally got one printed that came out good
<InPhase> That's getting the cpu fans some exercise.
<lf94> issue: I might kill my hand in a few hours :^)
<OlivierChafik[m]> what material did you print it with?
califax- has joined #openscad
califax has quit [Ping timeout: 276 seconds]
califax- is now known as califax
<InPhase> OlivierChafik[m]: This -DCGAL_DIR is not having adequate precedence or something. <CGAL/Polygon_mesh_processing/remesh_planar_patches.h> is coming up missing, although it is in cgal-sloriot-PMP-decimation/build/include/CGAL/
EkpyroticFrood has quit [Quit: So long, and thanks for all the fish.]
EkpyroticFrood has joined #openscad
<lf94> OlivierChafik[m]: PLA lol.
<lf94> that "stone" esque PLA
<lf94> At 0.2mm diameter and height
<lf94> I should really try TPU
<OlivierChafik[m]> hah nice. yeah TPU and I are not friends since I had to scrape it off my printbed (many times)
<OlivierChafik[m]> InPhase: ah yeah I actually uninstalled my homebrew CGAL
<InPhase> I've had great results with tpu.
<InPhase> OlivierChafik[m]: I tried a few modifications but I don't see it. I also don't see CGAL_DIR getting used anywhere.
<InPhase> Unless it's food for the find_package
<OlivierChafik[m]> InPhase: on my system it was tripping because if found the old cmakefile from CGAL in my /usr/local/include
<OlivierChafik[m]> what error are you getting?
<OlivierChafik[m]> CGAL_DIR seems to be a magic cmake variable.
<OlivierChafik[m]> (which it tries if all else fails, e.g. no trace of cgal whatsoever)
<OlivierChafik[m]> (sorry I meant it found it in /usr/local/lib/cmake
<OlivierChafik[m]> (I just savagely deleted all traces of cgal then handed CGAL_DIR to cmake as it prompted me to)
<InPhase> I do have a system CGAL, although I don't want to wipe that out. That's for my baseline builds.
<InPhase> Hmm. When I temporarily rename my system CGAL though, it finds no CGAL.
<InPhase> So this CGAL_DIR definitely does not work that way.
<OlivierChafik[m]> mmh I wonder if it has some sticky / cached variables somewhere
<OlivierChafik[m]> maybe it remembers where it found it previously or something
<OlivierChafik[m]> I did these instructions on a freshly cloned repo
<OlivierChafik[m]> cmake is still so mysterious to me
<InPhase> Alright, I think I'm officially tired of it. :) I did a ton of frustrating debugging during the work day already, and cmake is being wonky and ignoring attempts to get it to read a different .cmake file.
<InPhase> (There even IS a different cmake file in that directory, but it won't look at it.)
<InPhase> Oh hey! The render finished, because it took too long to build the replacement. lol
<InPhase> From 4 minutes to 1hr 26 minutes by enabling trust fast-csg.
<OlivierChafik[m]> yes build/CMakeCache.txt has a CGAL_DIR variable, so maybe after renaming your system dir it still thought it knew where it was and didn't run the same logic maybe
<OlivierChafik[m]> hahaha
<OlivierChafik[m]> so sorry for your loss (of time) :-(
<InPhase> A worthwhile strugle, I just used up all my oomph for build issues.
<OlivierChafik[m]> yeah I'm gonna try and get some nap time to do work tomorrow :-)
<InPhase> So I guess I've learned fast-csg is going pretty well and stable, but trust-fast-csg is a bit buggy yet.
<OlivierChafik[m]> hehe yeah trusting is a difficult business
<InPhase> My first several tests with trust-fast-csg were also not faster. They were all basically the same runtime (until this one).
<InPhase> It probably depends on the model though.
<OlivierChafik[m]> oh, interesting! yeah I find it often faster here, but there's so much variety. Like models that are astronomically slower except if you enable fast-csg-exact
<OlivierChafik[m]> I'm actually looking at our existing minkowski code rn, realized it's using Epick and not forcing it to exact anywhere. Probably a dead end, but then I'll sleep haha
<InPhase> Oh. I didn't even test that one because I assumed it would be slower.
<OlivierChafik[m]> It adds a bit of overhead but magically solves at least one terribly slow model that was posted on the original PR
<InPhase> There's always that one pathological case. :)
<OlivierChafik[m]> Ideally we'd want a non-lazy kernel but the CGAL folks will need a bit of nudging https://github.com/CGAL/cgal/pull/4495
<OlivierChafik[m]> there's also a fast-csg-exact-callback version that is hooked up with the corefinement visitor callbacks, but has more overhead. I'm now testing https://github.com/openscad/openscad/pull/4108 to see if that overhead is reduced (CGAL 5.4 only tho)
<InPhase> Nice.
<gbruno> [github] rcolyer opened issue #4109 (fast-csg eats fabric). https://github.com/openscad/openscad/issues/4109
<InPhase> OlivierChafik[m]: ^ A disappearance case for you.
<InPhase> OlivierChafik[m]: It's an interesting case as it's purely non-intersecting polyhedrons woven between each other.
<InPhase> And somehow that resulted in no output.
<InPhase> (No rush. Enjoy your nap time.)
<OlivierChafik[m]> InPhase: haha, thanks! And I'm almost happy with my debug logs / diagnostics refactoring but not quite. Off to sleep, 'night!
r3b has joined #openscad
<r3b> rant: the sad excuse for a pc fan that has been featured on the openscad.org page for as long as i can remember is just awful and needs to be retired.. it might technically work but... look at those fan blades! it's like "welcome to openscad, where you can make hacky potato approximations of things"
ur5us has quit [Ping timeout: 256 seconds]
<r3b> apologies for disturbing the peace
<InPhase> lol. Yeah, I've thought the same. :)
<InPhase> That and OpenSCAD doesn't look like that anymore.
<InPhase> We should probably pick some nice representative design. and get a modern screenshot.
<r3b> yeah, good plan
<r3b> wow, that felt good. i have hated that fan for years lol
<InPhase> Using something mechanical is probably a reasonable choice. That is a core strength.
<InPhase> Well I learned to stop looking at it, so I kind of forgot about it.
<r3b> yeah
ur5us has joined #openscad
ur5us has quit [Ping timeout: 240 seconds]
ur5us has joined #openscad
arebil has joined #openscad
<InPhase> teepee: From your lofty chair, did you spot any master branch changes that would radically shrink the font size of the Console window?
GNUmoon has quit [Ping timeout: 276 seconds]
<InPhase> I figured maybe we take some candidate screenshots and decide what we like. But then right when I went to do it I noticed how tiny the font had gotten in the master branch build I just did. https://imgur.com/a/jojQByf
teepee has quit [Ping timeout: 276 seconds]
teepee has joined #openscad
GNUmoon has joined #openscad
arebil has quit [Quit: Bye]
arebil has joined #openscad
ferdna has quit [Quit: Leaving]
J22 has joined #openscad
ur5us has quit [Ping timeout: 240 seconds]
GNUmoon has quit [Ping timeout: 276 seconds]
GNUmoon has joined #openscad
use-value has quit [Quit: use-value]
use-value has joined #openscad
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
J22 has quit [Quit: Client closed]
GNUmoon2 has joined #openscad
GNUmoon has quit [Ping timeout: 276 seconds]
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
J22 has joined #openscad
lastrodamo has joined #openscad
<Scopeuk> InPhase that looks (in the screen shot) about in line with the last release version on my machine
J22 has quit [Quit: Client closed]
arebil has joined #openscad
<peepsalot> InPhase: console font preferences were added shortly after the last release, you can set them under Preferences -> Advanced. I tried to set to a reasonable default, but it was difficult to determine the exact same font + size that was used previously.
<peepsalot> Also you can quickly change the apparent size with Ctrl-mousewheel, (that way of scaling does not keep over multiple sessions though)
metasean[m] has joined #openscad
<InPhase> peepsalot: Ah. I had tried Edit, Increase Font Size, but that only impacted the editor.
<peepsalot> and now its time for me to sleep... i'm 100% flipped around lately o_O
<InPhase> It seems that got set to "6" for me, which is indeed unusually small. I think these only recently got unusually small, but maybe I misclicked something in preferences while swooshing around.
<peepsalot> its possible that it got changed while scrolling by. i don't know if it can be disabled, but qt will count hover as focus, and change select boxes while you scrollwheel through preferences
<peepsalot> its bitten me a few times, kinda annoying
<InPhase> Font size auto-determination can also be tricky, particularly if it was a method that did not factor in OS dpi settings. (The scale of the smallness was about on par with my DPI deviation from standard.)
J22 has joined #openscad
<J22> TGIF
<Scopeuk> for some reason visual studio + cmake
<Scopeuk> +vcpkg always feels a little bit like balancing two jack russels on a surfboard ontop of a fence
<Scopeuk> when it works its impressive but there is no telling when that will be
<OlivierChafik[m]> InPhase: re/ yesterday's build errors, the CGAL_DIR needed to be passed as an absolute path somehow, e.g. -DCGAL_DIR=$PWD/../../cgal/build
Alicia has quit [Ping timeout: 260 seconds]
Alicia has joined #openscad
<InPhase> OlivierChafik[m]: I tried that too, it also didn't work.
<InPhase> That was my 4th attempt last night. :)
<teepee> If CGAL is installed, I think CGAL_DIR needs to point to the cmake file?
<OlivierChafik[m]> to the folder where the cmake file is
<InPhase> I can give it another go tonight.
<teepee> right, the DIR :)
<OlivierChafik[m]> 😅
<InPhase> If it continues to fail I'll try DUR.
<teepee> C_DUR?
<OlivierChafik[m]> DOH
<OlivierChafik[m]> DUH
<InPhase> JUST_WORK=true
<ecraven> hm.. now I'm starting to design things in openscad that the 3d printer can't print :/
<J22> ecraven wrong printer - Ü
<InPhase> Well that can be fun too. :) But don't do that for things you want to print.
<ecraven> well, I wanted lots of 0.5mm holes :P
<ecraven> for a laminar air assist nozzle for the laser cutter
<InPhase> You can make 0.5mm holes. Render them as 0.8mm or 0.9mm
<InPhase> Design it in as nominal + tolerance
<J22> that is possible but you need to tune your printer (horizontal expansion) or  make the holes bigger so they are .5 when printed
<ecraven> hehe, I don't know much about 3d-printing, it's not even my printer, I'm having a friend print it.. I'll try the 1mm holes, and see how they work out
<J22> cura also has an experimental small feature setting
<InPhase> cura will also typically ignore structures below twice the extruder size now.
<ecraven> no idea if this is even a good design: https://imgur.com/a/L15cEOP
<InPhase> Ah, well I didn't know there was a new setting, but that will help.
<ecraven> the friend uses prusa slicer
<InPhase> Why small holes there instead of larger ones?
<ecraven> to equalize the airflow, I'm assuming many smaller holes are better than fewer larger ones
<ecraven> but I'm no engineer, so this is just guesswork ;)
<J22> inPhase  that feature seems to just slow down the printer
<J22> ecraven  .. this is not how this works
<ecraven> ;) I'd be happy about any tips or ideas on how to make this better
<J22> without the nozzle orifice you could get laminar flow   but  i assume you get problems with the space between the holes
<InPhase> ecraven: Many small holes act as a restrictor. Which is sometimes useful, when you want less flow.
<InPhase> If it's pressurized flow many small holes also increase the fluid (air is a fluid) exit velocity, but still restrict total flow.
<J22> calculate the nozzle area and  make 1mm holes  so their area is 1.5-2× that area
<J22> instead of holes you could use fins to remove any rotation
<InPhase> Water-saver shower heads for example are based on this many small holes principle, giving you high speed jets using less water that pretend to clean you just as well as if you're a dirty dish needing water torch scrubbing.
<ecraven> it's 2mm diameter, so that'd make it about pi mm² in area. so I should create about 2 pi area holes in total?
<ecraven> that'd be about 8 holes. I have many more
<ecraven> about 4 times as many, I think ;)
<J22> i got 4 holes each 1mm
<ecraven> this is what I had before, https://imgur.com/a/TjipWXR if the "laminar" thing I'm trying to do is not going to perform any better, I might as well stick with that ;)
<J22> sorry  yes twice that  or  6 holes    ( it is just bit eyeballing to cover the higher surface friction )
<ecraven> 0.5²×π ≅ 0.79, so 2×π÷0.79 is about 8?
<ecraven> hm.. so if I have too many, that's bad too, right? then I'll get turbulences in the chamber?
<J22> it is hard to predict  as  there are many factors
<ecraven> will there even be an actual difference in use?
<J22> the question is  if there is a pressure difference between the holes
<ecraven> also, openscad is great! it really clicks for me to design things (still very badly :-/) in openscad's way
<J22> you don't want any rotation  - laminar flow is nice but probably not here
<ecraven> why do I not want rotation?
<ecraven> in the end, it "just" has to blow steadily out the top nozzle, in a fine stream, in order to blow *into* the laser's kerf
<J22> if you have slight rotation this will increase when pushed in the middle  - then the centripetal force will rip the stream wide open .. but you want a stream focused on the  laser focus point
<J22> ( also the flow is used to cool the lens - there rotation would be beneficial
<J22> )
<ecraven> hm.. I could add additional channels going towards the "bottom"
<ecraven> this is all a lot of guesswork on my part, I have no idea how these things actually work ;) it's "only" a 5.5W optical output diode laser, so maybe it doesn't matter much anyway
<J22> if this is a laser cutter .. keep in mind the laser will be cone shaped - it seems your channel blocks the laser
<ecraven> which part of the channel? the focus point is about 2cm *above* the nozzle, so it should be fine?
<ecraven> the through-channel (and the exit of the nozzle at the top) is 2mm, the laser is 0.1mm, I think, if not even thinner?
<J22> keep a free area from the lens diameter to the nozzle diameter
<J22> the laser focus  in a .1 mm spot  so it the lens it is  more like 1-2 mm
epony has quit [Read error: Connection reset by peer]
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
J2248 has joined #openscad
<J2248> ecraven  i would think about some coaxial design
J22 has quit [Ping timeout: 256 seconds]
<J2248> for 5w this is probably  not that importand ..  the flow might cool so your results get worse
arebil has joined #openscad
<ecraven> J2248: coaxial as in "the air flow goes beside the laser beam, but not in the same channel"?
<J2248> yes   like a ring cone  / segmented ring slit
<ecraven> all the way to the nozzle? with angled exits, so the streams go towards the focus point?
<J2248> so you can have high velocity without needing a lot of air .. so  it doesn't cool the spot to much
<J2248> or  like  having 3-5 nozzles around the laser
<ecraven> ok, I'll do more measurements, and think about that, but that does sound like a better design.. thanks a lot for the explanations and patience!
<J2248> always welcome
<ecraven> is it ok to have a torus to "distribute" the air into the other channels?
<J2248> ecraven:   something like this https://pasteboard.co/0vNJQnOiRRJy.png
<J2248> ( that would be the object  which is differenced from the nozzle body to create the channels .. a cone in the middle for the laser need to be removed too )
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
epony has joined #openscad
epony has quit [Ping timeout: 240 seconds]
<ecraven> yea, that's what I was thinking. thanks very much!
Foidi[m] has quit [Quit: You have been kicked for being idle]
epony has joined #openscad
epony has quit [Ping timeout: 240 seconds]
<ecraven> something like this? https://imgur.com/a/XRfOfEH
<ecraven> so the sum of my nozzle exit areas should be about the same as the smallest area of the inlet (from the right), correct?
<ecraven> or maybe 1.5-2x that, to account for friction?
<ecraven> hm.. what if I go from the inlet to the "bottom" of the nozzle, into the wide laser canal? won't the airflow then be mostly laminar anyway, because there's only one way "out"?
<ecraven> sorry, not bottom of the nozzle, but the part closest to the lens?
<InPhase> What's driving this air flow?
<ecraven> right now an aquarium pump (this one https://www.amazon.de/gp/product/B00BNSQ16K/ref=ppx_yo_dt_b_asin_title_o04_s00?ie=UTF8&psc=1), but I want to attach a valve and maybe drive it from an actual compressor, but with a regulator so I can go to ~10-15 psi (that's what I read might be good pressure)
epony has joined #openscad
<J2248> ecraven the exit area need to be same or smaller than the inlet .. you wand the else  the inlet become a bottleneck
<J2248> also instead of  constant diameter .. make the channel as big as possible at the base
<J2248> as  a tube will loose pressure due to its length
<ecraven> what about something like this? https://imgur.com/a/zNRk3A8
<ecraven> (the actual inlet is missing, but maybe you understand what I mean)
<J2248> you can do this .. but  use bigger holes or just open the hole torus ..   you want the  nozzle to be the smallest point
<J2248> having nozzles beside the laser channel  allows you to make them narrower than  the laser need to pass .. to minimize volume flow
<J2248> for the bigger laser the airflow is needed to cool the lens and  remove material from the kerf  but  should be as minimal as possible as the cooling in the kerf is negative for cutting
<J2248> e.g.  when cutting wood  - the more air the darker the wood gets as it is cooled and burns instead of vaporize
<J2248> in your last image the inlet into the central cone is radial .. so  air will meet at the center and likely turns into rotation,  so  an angle closer to axial is beneficial
<ecraven> hm.. making the ducts into the central laser cone "helical" (so that the air actually spins around) isn't good either, right?
<ecraven> just so that the streams wouldn't hit each other
rawgreaze has quit [Quit: ZNC 1.8.2 - https://znc.in]
<J2248> here an image what happens if water rotates through a small orifice .. https://pasteboard.co/5OtQerNcMRfA.jpg
<J2248> you can do that .. but need to stop the rotation when leaving the funnel ..  by converting the rotation into axial flow you also gain energy so it speeds up
<ecraven> hm.. so helical ducts on the "bottom" and then fins on the "top"? maybe I shouldn't try to design something I have no idea about :P :D
<J2248> here is both  a central axial flow with a rotating  outer shield https://pasteboard.co/7Cqtjszs7o3n.jpg
<J2248>  you will learn most if you try things you have no clue  ..  just use water  (shower hose) to test if it is doing what you think
<ecraven> I was just thinking that with a purely coaxial design that only exits at the tip of the nozzle, smoke might stay *inside* the main canal and "dilute" (what's the proper english word here?) the laser
<J2248> inside the nozzle is no smoke  and outside is so much turbulence that this will mix .. also  a fast stream will pull surrounding air with   (Bernoulli)
<J2248> ecraven:  diffuse scatter or absorb  is probably what you meant
EkpyroticFrood has quit [Quit: So long, and thanks for all the fish.]
EkpyroticFrood has joined #openscad
EkpyroticFrood has quit [Read error: Connection reset by peer]
EkpyroticFrood has joined #openscad
arebil has joined #openscad
operhiem1 has quit [Quit: The Lounge - https://thelounge.chat]
operhiem1 has joined #openscad
ur5us has joined #openscad
<Scopeuk> Going back a bit J2248 but for .5mm holes just print solid and drill, maybe print a grid pattern to help you locate the holes
<InPhase> -1 points for subtractive manufacturing methods. ;)
ur5us has quit [Remote host closed the connection]
ur5us has joined #openscad
epony has quit [Ping timeout: 240 seconds]
<lf94> lol
epony has joined #openscad
califax has quit [Ping timeout: 276 seconds]
califax has joined #openscad
Guest46 has joined #openscad
Guest46 has quit [Client Quit]
<Scopeuk> Its not subtractive is negative additive :p
<InPhase> :)
epony has quit [Ping timeout: 240 seconds]
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
<J2248> Scopeuk i would agree (and do this my self for all holes that need to be exact ⇒ using a reamer  )  but  try this here in his example https://imgur.com/a/L15cEOP
<Scopeuk> Ahh, that looks almost like a messed "air multiplier"
<Scopeuk> Seen so many of them cooling for f1 drivers in the pits
ur5us has quit [Ping timeout: 240 seconds]
<J2248> years ago (before dyson stole this)  i had the idea to use that ring "air multiplier" as  jet drive for ships
<J2248> (inside a hydrofoil)
<Scopeuk> My understanding is all dyson have is sticking a fan in the base, fluid through opening to create drag on surrounding fluid is steam machinery tech, its how the steam to vacuum adaptors work
<Scopeuk> https://www.gutenberg.org/ebooks/28553 is great fun for that sort of stuff
<J2248> sure but some patent are on that ..  like on most trivial things - they are trivial after you had the idea to use physics that way - Ü
<J2248> nice book !
<Scopeuk> Yeh annoyingly to invalidate a patent on prior art you have to prove all terms in one item. Hence all the with a computer nonsense etc
epony has joined #openscad
<teepee> OlivierChafik[m]: did you intend to post all the new PRs as draft?
epony has quit [Ping timeout: 240 seconds]
J2248 has quit [Quit: Client closed]
J2248 has joined #openscad
ur5us has joined #openscad
ur5us_ has joined #openscad
ur5us has quit [Read error: Connection reset by peer]
myosotis has joined #openscad
<OlivierChafik[m]> teepee: I did haha! Did lots of git gymnastics to unentangle lots of local changes and didn't take the time to run all the tests separately, counting on CI for that ✌️
<OlivierChafik[m]> I've rolled my sleeves and I'm now converting all the minkowski code to hybrid + surface_mesh (+ a couple of other tweaks)
<teepee> no worries, just asking because I've had the case myself that github remembers the state of the merge button
<OlivierChafik[m]> teepee: ok two are ready to review now, including the one for InPhase (less verbose logging)
lastrodamo has quit [Quit: Leaving]
<teepee> for the remeshing, I guess we want to wait till the PR has landed in CGAL repo
<teepee> if that happend, we could grab that as patch
<OlivierChafik[m]> yeah up to you, that or we submit these couple of lines of dead code to make it easier for us to patch in his branch. TBH this is such a huge fix that I'm tempted to fork CGAL and have a 5.future branch
<OlivierChafik[m]> (and convince you to build from that haha)
<OlivierChafik[m]> or if only we could maybe just get the bits from his PR that work independently
<teepee> I would need a patch against the release source bundle anyway, so there's not much point of a separate repo
<teepee> just want to make sure it's officially accepted, as currently it has the tag "not accepted yet"
<OlivierChafik[m]> hah yes fair enough 👌
epony has joined #openscad
<gbruno> [github] t-paul pushed 4 modifications (Merge pull request #4110 from ochafik/fast-csg-debug
<gbruno> [github] t-paul pushed 1 modifications (Merge pull request #4112 from ochafik/hull-points-vector
<ndnihil> soon
<ndnihil> soon I will have -allll- the closing parens
<ndnihil> all mine
<teepee> oh noooo!
<ndnihil> sorry, saw your message the other day but was tied up
<teepee> poor lisp lovers
<ndnihil> nfi what that's all about, will have a look
Netfool has joined #openscad
<ndnihil> did github change their syntax or something?
<ndnihil> closing paren and appended url are still in that string
<ndnihil> but it's not showing either
gsker has joined #openscad
<gsker> Hi! I'm running openscad nightly and it identifies itself as version 2022.02.10.nightly (git eab434c). Is that a github commit? I have tried finding it at https://github.com/openscad/openscad and can not.
<gsker> My reason for looking for the commit was to figure out if I was succeeding in enabling fast-csg. It turns out that putting --enable fast-csg at the commandline was not doing it -- But setting it in the Preferences dialog did. Hmmm.
ur5us_ has quit [Remote host closed the connection]
<teepee> yes, that's the github commit
<OlivierChafik[m]> Oh, weird, I'm mostly using it in the command line. Could you be having two versions installed maybe and a stable one in the path? Which os are you on?
ur5us has joined #openscad
<gsker> Nope. Debian testing
<OlivierChafik[m]> The command line should warn you if it doesn't recognize a feature (e.g. try --enable=foo)
<gsker> I run it as /usr/bin/openscad-nightly. I'm gonna try removing the settings file entirely and seeting what happens then.
<gsker> gsker:[master]/tmp/openscad/scad-utils>openscad-nightly --enable foo
<gsker> WARNING: Ignoring request to enable unknown feature 'foo'.
<OlivierChafik[m]> then does it also warn against --enable=fast-csg ?
<teepee> what's --help showing?
<gsker> help shows all the settings available.
<gsker> --enable arg enable experimental features (specify 'all' for
<gsker> enabling all available features): fast-csg |
<gsker> fast-csg-exact-callbacks | roof |
<gsker> fast-csg-debug-corefinement | fast-csg-exact |
<gsker> fast-csg-trust-corefinement |
<gsker> input-driver-dbus | lazy-union |
<gsker> vertex-object-renderers |
<gsker> vertex-object-renderers-indexing |
<gsker> vertex-object-renderers-direct |
<gsker> vertex-object-renderers-prealloc | textmetrics |
<gsker> import-function
<OlivierChafik[m]> hah there you go so --enable=fast-csg should just work as intended (well, unless you help find more bugs ;-))
<gsker> That's the whole point. When I ran with --enable=fast-csg it does not enable it.
<gsker> But when I change it in the Preferences dialog it DEFINITELY works.
snaked has quit [Remote host closed the connection]
snaked has joined #openscad
<OlivierChafik[m]> you mean you see the fast-csg logs in the UI console and not in the command line's stdout?
<gsker> As a test I deleted the OpenSCAD.conf file and started with the command line option. It does not propogate to the UI.
<teepee> yes, that's normal
<gsker> I never run the pure commandline. I'm always in the UI.
<gsker> Normal. Ooooooh. Okay.
<gsker> (NOT sarcastic, btw) I get it.
<OlivierChafik[m]> ah ok, yes --enable only enables it for batch jobs (i.e. if you specify -o output.stl in the command line), it doesn't set the UI setting
<dalias> is there an existing openscad module for making printable worm gears?
<gsker> well I would not expect -o to do that, but I would expect actual options to override the UI. But since it isn't intended to I'll just pick up the shattered pieces of my worthless life and move on.
<gsker> LOL
<teepee> freecad for example runs openscad command line
<teepee> when importing files, I don't think it would be nice if that would also always switching options around
<teepee> also it would be difficult to handle when running concurrent instances
<gsker> I'm not sure what you are saying. If I run firefox with no options it picks up its options from the config. But if I include an option, then it's an option only for that one run.
<gsker> That's sorta the flow my brain had in mind.
<J2248> dalias  evolute gears?
<dalias> not sure what that means. i mean a worm gear and worm wheel
<dalias> worm gear is simple helix of a thread profile, so easy to do
<dalias> but i dont know how to do the wheel
<J2248> there are very different worm gears
<gsker> Since my original question was to point to the source for my nightly could someone find the github location for my build? That should give me a clue as to how to find it in the future.
<dalias> aside from just dividing up time finely and differencing the worm from the wheel at each step, which is actually a plausible way to do it :-p
ferdna has joined #openscad
<gsker> 2022.02.10.nightly (git eab434c)
<gsker> is it in the master branch?
<teepee> yes
<J2248> dalias also a worm is not a helix of a thread but the helix of a gear
<gsker> Thanks!
<OlivierChafik[m]> gsker: I agree a per-run setting would make sense, but then when the user opens the settings we'd have to give a hint as to which settings have been overridden for the current run and which are set persistently. Or we could just throw an error when any --enable flag is passed in a GUI invocation
<dalias> my motivation is that the guy behind flex3drive (one-man operation) is kinda missing and i want to determine if spare parts are printable
<gsker> Yup. I would not spend any time adding that ability if I were you.
<dalias> j2248, stuff i'd read on worms in the past suggested the profile was more like a thread in some ways. but in any case it's a profile you can produce as a helix from a 2d profile
<gsker> Thanks for the help!
<OlivierChafik[m]> np, happy rendering!
myosotis has quit [Remote host closed the connection]
<J2248> dalias yes the worm is a thread  - maybe i misunderstood what you meant with extruding a thread profile, the profile is a gear with Evolute teeth
<dalias> j2248, the worm in that collection doesn't seem to have a fancy wheel that matches the worm curvature, just flat
<J2248> should be helical cut
<dalias> yes it's angled as a helical gear
<J2248> what you search is called a globoid
<dalias> yeah that sounds right
<dalias> well the one i'm looking to be able to clone is single-enveloping not double-enveloping
<dalias> the worm part is just a straight helix but the wheel is enveloping of the worm
<dalias> watched a nice video today on how to machine them :-) but the whole idea is to have high power in low mass, so metal isn't really a solution even if i had a lathe :)
<dalias> i suppose if i had a lathe, the ideal way to make them would be to machine from a chunk of POM :)
<J2248> or print POM ..   the globoid cant't be printed  without supports so cylindrical worms are easier
<dalias> yeah... POM is my nemesis for printing :-p
<dalias> and any warping here would be a non-starter
<J2248> i think removing a torus  (with the core diameter of the worm )from a helical gear could be a good approximation for a worm gear wheel
<J2248> POM also just doesn't  adhere to anything except itself .. i am using POM sheets to print on
use-value has quit [Ping timeout: 240 seconds]
use-value has joined #openscad
ur5us has quit [Ping timeout: 256 seconds]
<dalias> it actually adheres to suitably rough paper
<dalias> however last time i built a surface of suitable paper glued to my bed...
<dalias> the warping pom actually ripped up the glued corners and made the paper curl :-p
<dalias> "drywall tape" is the recommended paper type, btw
<dalias> so anyway i think i could make it work with a suitably rigid bed material to glue paper permanently to (and sand down and put new paper on top of when it gets old
<dalias> but i didn't want to further ruin my buildtak-clone :/
<dalias> and i didn't want to keep playing with bed leveling knobs
<dalias> thinking about adapting my endstop to be against the top of the bed, then swapping out bed surfaces will be much less painful