<J24k96>
i can't reproduce the crash but i am using 2024.01.03
teepee_ has joined #openscad
<kintel>
For a huge design, ideally try to slim it down before reporting, as it can be quite some work to get to the meat
teepee has quit [Ping timeout: 255 seconds]
teepee_ is now known as teepee
<J24k96>
yeah as i said i can't replicate the problem at all so not sure what went wrong
<teepee>
mmu_man: --debug takes a list of source code file names too
<teepee>
e.g. --debug import_off,fontcache
<J24k96>
kintel it crashed with 2024.02.04
<J24k96>
3mf export working fine
<J24k96>
well great and now it don't
<mmu_man>
teepee: yeah I used all but the GUI was already setup and I expected it in the terminal…
<teepee>
I've seen that, just an additional note as "all" can be annoying
<J24k96>
i mean it don't crashes anymore .. probably some of these odd windows behavior issues again
<mmu_man>
yeah it's verbose :)
<mmu_man>
now what's the target for tests…
<mmu_man>
ah ctest directly
<mmu_man>
thought the scale was the cause for some failed tests but seems not
<mmu_man>
probably some tests require CGAL and they won't work anyway
<mmu_man>
ohhh yeah ok, KiCad's default unit for WRL export is meter, no wonder meshlab exported to OFF as minuscule stuff
<mmu_man>
wrl seems to have a scale command to specify it… maybe I should add a wrl importer
J24k71 has joined #openscad
<mmu_man>
bad header line 'OFF 14 24 0'
<mmu_man>
hmm seems like the test files actually merged the first two lines?
J24k96 has quit [Ping timeout: 250 seconds]
<mmu_man>
the spec is not really clear on that
<mmu_man>
anyway, should sleep
omegatron has quit [Quit: Power is a curious thing. It can be contained, hidden, locked away, and yet it always breaks free.]
J24k71 has quit [Quit: Client closed]
<kintel>
peeps[work] In case you're around: I see you did some work on projection a while back
<kintel>
I'm trying to debug some Clipper issues, and I struggle to remember/understand how (no-cut) projection works
<kintel>
Basically, it looks like we do this (example starting from a convex PolySet):
<kintel>
1. remove z coordinate and convert the PolySet to Polygon2d. We convert both front-and back-facing polygons.
<kintel>
2. Perform a Clipper union with fillrule Non-Zero
<kintel>
However, my understanding is that when converting both front- and back-facing polygons, they will cancel each other out, resulting in just zero. Why does this work?
<peeps[zen]>
i don't recall making changes to projection specifically. are you sure it wasn't superficial refactoring
<peeps[zen]>
yeah not sure how/where/if it corrects for front/back facing. looks like I basically only split a larger function into smaller ones with copy/paste, didn't really dig into the inner workings
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
J24k has joined #openscad
L29Ah has quit [Ping timeout: 256 seconds]
L29Ah has joined #openscad
Bambukah has joined #openscad
<Bambukah>
Good morning :)
<Bambukah>
Just play with threads on BOSL2 and make a test... and it fails! Need to learn more thing about threading :)
Guest86 has joined #openscad
Guest86 has quit [Quit: Client closed]
mmu_man has joined #openscad
epony has joined #openscad
mmu_man has quit [Ping timeout: 252 seconds]
Bambukah has quit [Quit: Connection closed]
ToAruShiroiNeko has quit [Ping timeout: 260 seconds]
Guest97 has joined #openscad
To_Aru_Shiroi_Ne has joined #openscad
<Guest97>
Haven't seen irc in 15 years..
<J24k>
it has been there all the time
<Guest97>
I'm using someone elses script to basically generate a border of a letter
<Guest97>
difference(){
<Guest97>
linear_extrude(height=depth)
<Guest97>
minkowski(){
<Guest97>
text(letter,sized,style);
<Guest97>
circle(walls);
<Guest97>
}
<Guest97>
translate([0,0,base])
<Guest97>
linear_extrude(height=depth+2)
<Guest97>
text(letter,sized,style);
<Guest97>
}
<Guest97>
This basically cuts the inner(non circle) part of the shape and it works great, but now I'm trying to figure out how to cut the outside instead(I can't use a different size letter as it doesn't align up correctly, any hints?
<Guest97>
Thats pretty close, but not quite what I'm trying to get to, I'm trying to have the inner text, then empty space, then an outer border, the one you sent me does the same as the original
<Guest97>
That last link is great, but I still need to get the inner partt
<Guest97>
Inset seems to be the right track, is there a way to create the solid object, and reuse it, once to create the outshell and then once to do the inset
<J24k>
petaflot maybe you have no overlap - use the modifiier # * ! to show or mark objects
<J24k>
petaflot get an error that you mix 2D and 3D
<petaflot>
J24k: what do you mean by no overlap? there are some cylinders that are half inside the cubic base, they should definitely show. and as I mentionned, preview is fine but render is not
<petaflot>
the only thing 2D is text that went through linear_extrude
<petaflot>
got any doc for the # * ! modifier(s)? I've never used them before
<J24k>
I can see there are regions where is no overlap between discs
<petaflot>
yeah that was was it.. I had forgottent the extrusion there. thanks a lot
<J24k>
between some discs there is only a very thin wall or none maybe make thicker walls or overlap for no walls
<J24k>
And i would use cylinder(); and not this strange library
<petaflot>
J24k: I love my library
<petaflot>
the only place I use cylinder is when I need to draw cones
<petaflot>
wall thickeness are exactly as intended (I use a pythong script to generate the scad file)
<petaflot>
the thing I like most about my "prism()" primitive is that centering (or not) objects like cubes is way easier
<J24k>
cylinder (h=5,d=20) giving you a normal cylinder .. the problem is that your formatting is not ideal and your library is .. lets say if you want support you make it very difficult besides using French
<petaflot>
.. on whatever axis you want. by default it acts like the cylinder, centered on X and Y
mmu_man has quit [Ping timeout: 268 seconds]
<petaflot>
I've been using openscad for 10+ years... pretty much the first time I need support (and the first time I use text in a design)
<J24k>
You can absolutely do your own things - it is just super difficult for others to understand your code.
<petaflot>
fair enough
<J24k>
In general it helps to make a small simple version so it is easier to find the issue
<teepee>
mmu_man: did you rebase to or sync with latest master?
<teepee>
I think you branched of when there was some build issue
<teepee>
ah, I see force-poshed, cool, crossing fingers :)
<mmu_man>
yes
<teepee>
it might be useful to make the error string passed to AsciiError unique
<mmu_man>
you mean the dups? compiler should be able to optimize that away
<teepee>
I mean have different text so it's easier to reference which exact instance did produce the error, I'm not sure the other information always is enough to determine that
<teepee>
like a user reports "bad header"
<mmu_man>
hmm there's lineno but I'm not sure I increment it correctly