L29Ah has quit [Remote host closed the connection]
L29Ah has joined #solvespace
_whitelogger has joined #solvespace
cr1901_modern has joined #solvespace
balrog has quit [Ping timeout: 240 seconds]
balrog has joined #solvespace
L29Ah has quit [Remote host closed the connection]
L29Ah has joined #solvespace
<whitequark>
lexszero: having used solvespace for real-world mechanical engineering tasks, I beg to differ :p
<whitequark>
it does miss some of the fancier features from large cad packages, that's true
<whitequark>
wpwrak: keep in mind that drugs affect even animals only slightly different from us in a dramatically different ways
<whitequark>
I'd really like to see webs made by spiders dosed with a set of substances that are not psychoactive in humans
<whitequark>
as it is the article is basically just yellow press... and if you look at the comments, you see people assuming that the findings translate in any way whatsoever to humans
<whitequark>
try giving caffeine to your dog (don't unless you hate it a lot).
<lexszero>
it largely depends on the tasks itself. i'm too used it succesfully for _some_ tasks i have.
<whitequark>
lexszero: right now the high-level focus is on getting more useful highly nested parametricity
<whitequark>
so things like screws or extrusions; you can already use those but it's too much work for any sizable assembluy
<el3phanten>
I have used freecad for a while. And recently tried solvespace. What I think is nice here, is that it is so much easier to create the objects I need. So if I need extrusions and all that I can use freecad for that
<el3phanten>
one does not replace the other. So to use them together is a good way to work for me
<lexszero>
when i tried to use freecad ~1.5y ago it was totally unusable because of sudden crashes and overall sluggishness
<el3phanten>
lexszero: ok, they make improvements all the time. I experience crashes sometimes too tho
<whitequark>
I started maintaining solvespace after trying to use freecad for two weeks
<whitequark>
freecad is a piece of shit
<whitequark>
they still don't have a working assembly workbench do they?
<whitequark>
as for crashes, yeah, solvespace crashes too... what it *never* does is corrupts your savefile after a crash
<whitequark>
which freecad did at least twice just in the short span where I tried to use it
<lexszero>
hmm, afair it was possible to do say "this plane of part A should be on that plane of part B"
<whitequark>
I don't think it was
<el3phanten>
I did not say freecad was better or a replacement. But there are fillets and all that fancy stuff
<el3phanten>
and I used t to make some complete drawings
<el3phanten>
as long projects are opensource I wont complain. I could just throw money at autodesk if I want to complain
<el3phanten>
if you know what I mean
<el3phanten>
opensource, you can just start committing :)
<whitequark>
oh, I didn't take it like that at all
<whitequark>
it's just that I'm still bitter about the time I wasted on freecad, and how much time I spent wrongly thinking there's no usable FOSS CAD
<whitequark>
as for fillets: yes. though through no virtue of freecad on its own. all the hard work was done by opencascade.
<el3phanten>
whitequark an lexszero: do you use any other opensource cad programs for stuff like that? extrution along path, fillets etc..
<el3phanten>
I just know about freecad, thats why I mentioned it
<whitequark>
no, I design to avoid it :)
<whitequark>
but I'll be in a real bind if I really *need* it
<el3phanten>
ok :)
<whitequark>
keep in mind we do want lofting etc and there are vaguely specific plans to do it
<lexszero>
el3phanten: openscad
<whitequark>
evil-spirit is studying NURBS stuff
<whitequark>
openscad doesn't help here because you can't import triangle meshes into solvespace
<whitequark>
importing step probably makes more sense
<el3phanten>
there is a openscad workbench in freecad if that helps :D
<el3phanten>
but whitequark nice to hear that this is getting bigger
<el3phanten>
I like the workflow in solvespace much better than freecad
<whitequark>
the openscad workbench in freecad is mostly opencascade too fwiw
<el3phanten>
yea I have not used openscad much.
<bcoburn>
serious professional cad packages crash _constantly_. my coworker used to keep a tally on a little whiteboard above his desk, got 5-6 crashes in one morning once
<whitequark>
yeah, I've seen people whose day job involves CAD salivate at solvespace for exactly this reason
<GitHub>
[solvespace] whitequark commented on issue #231: Unable to reproduce. Just set up a fresh build directory using Visual Studio 14 2015 on Windows 10 x64, and the build succeeded. I'm not sure what I can do in regards to this report, so closing.... https://github.com/solvespace/solvespace/issues/231#issuecomment-292563405
<GitHub>
[solvespace] whitequark commented on issue #231: Unable to reproduce. Just set up a fresh build directory using Visual Studio 14 2015 on Windows 10 x64, and the build succeeded. I'm not sure what I can do in regards to this report, so closing.... https://github.com/solvespace/solvespace/issues/231#issuecomment-292563405
<GitHub>
[solvespace] whitequark commented on issue #231: Unable to reproduce. Just set up a fresh build directory using Visual Studio 14 2015 on Windows 10 x64 using the 64-bit tools, and the build succeeded. I'm not sure what I can do in regards to this report, so closing.... https://github.com/solvespace/solvespace/issues/231#issuecomment-292563405
<GitHub>
[solvespace] whitequark commented on issue #232: This is expected, and was extensively discussed before. SolveSpace cannot be built with Debian Jessie (stable, soon to be oldstable) but can be built with Debian Stretch (testing, soon to be stable). https://github.com/solvespace/solvespace/issues/232#issuecomment-292577663
<Jonimus>
whitequark: yeah solvespace's stability is part of what makes it great, and I have to say the show triangles option and step export being rock solid have me sold.
L29Ah has quit [Remote host closed the connection]
<whitequark>
wait, *those* are the features that you care about most?
<whitequark>
why??
L29Ah has joined #solvespace
<Jonimus>
I guess I just like show triangles fo understanding whats going on, and since I work with CAM software haveing good step export makes actually using the models easy.
<whitequark>
should solvespace even gain IGES export?
<whitequark>
is that useful?
<Jonimus>
There are quite a few CAM packages that doesn't support STEP import, or it costs extra so it may be. IGES being the older format is the main reason its seems to be better supported, though STEP is generally better when you can use it.
<whitequark>
I see
<whitequark>
personally I find STEP completely nuts
<Jonimus>
IGES files generally only have surface data where STEP generally has "solid models"
<whitequark>
have you looked into how it is structured internally?
<Jonimus>
A bit, IGES isn't much better IIRC.
<whitequark>
disgusting
<Jonimus>
Per the standards bodies STEP replaces IGES but they are in no way compatible.
<whitequark>
but it's about what I expected
<whitequark>
so hm
<whitequark>
pretty sure STEP also defines surfaces
<whitequark>
isn't a solid just a region of space bounded by surfaces?
<Jonimus>
IIRC a solid shares edge data where surface models each surface is independant.
<whitequark>
oh, like having a reference to the same edge data structure?
<Jonimus>
I believe so, don't quote me on that though.
<whitequark>
I see, that makes sense
<whitequark>
there was a complaint recently about our STEP export
<whitequark>
that it didn't define a solid
<Jonimus>
I guess I didn't try importing it into anything yet, I should do so and see what our software's translators say.
<Jonimus>
Side note, IGES can technically contain solids, I've only ever seen it once and I think most importers would choke on it.
<Jonimus>
Also there are a ton of CAD packages with utter shite IGES export, like none of them realize IGES can do circles so all the things that should be round end up being not quote round splines
<wpwrak>
i actually have an earlier version, much simpler (didn't have the battery contacts), which i did by scripting freecad. i kinda doubt i would have gotten that far with the gui.
<Jonimus>
That causes all kinds of havok on hole recognition for drilling among other issues.
<whitequark>
Jonimus: oh yeah I can see that
<whitequark>
aren't there packages out there that read G-code, approximate piecewise linear moves with bezier curves, optimize, and then make them back into pwl?
<whitequark>
so much utterly wasted effort
<Jonimus>
pwl?
<whitequark>
solvespace goes to a quite large amount of effort to preserve circles (and other exact representations) in dxf export
<whitequark>
piecewise linear
<Jonimus>
oh gotcha, yeah I think that is mostly to compensate for shitty lookahead in NC controls.
<Jonimus>
and yes I liek to use ATM Machines.
<whitequark>
so I guess I wouldn't spend much effort supporting IGES
<whitequark>
actually I don't think anyone has wanted it yet, which speaks volumes to its usefulness
<Jonimus>
Importing would be more important than export.
<whitequark>
we don't even have STEP import yet..
<whitequark>
... and for IGES import we'll need some library, so once that's done export is a piece of cake most likely
<Jonimus>
though getting step Import would be huge, plus then you can test round triping models and seeing what errors creep in.
<whitequark>
it wouldn't be true roundtripping
<wpwrak>
wish solvespace had properly working STL export from the command line. being able to automate some workflows would be great.
<whitequark>
wpwrak: what's the current issue?
<wpwrak>
whitequark: that crash when trying to print a complaint. and -t being unsupported
<wpwrak>
so, old issues. or did you fix any of that recently ? haven't updated for a while. i have a big stack of vital improvements in the version i use, so stopped updating once all the basic works.
<whitequark>
the former will be fixed soon as a side effect of something else
<wpwrak>
yes, but i think it only applies to NURBS, no ? while meshes take the display chord ?
<wpwrak>
and NURBS are basically useless for me. they always fail when things get even slightly complicated, and then i have to proceed with a mesh.
<whitequark>
um, no
<whitequark>
sec
<whitequark>
let me verify
<whitequark>
>SS.exportChordTol = chordTol;
<whitequark>
so... export-mesh uses --chord-tol for export chord tolerance
<whitequark>
and non-export commands set SS.chordTol
<wpwrak>
hmm yes, ran a few tests and it looks good. sorry for the false alarm, and thanks for fixing it ! :)
<whitequark>
fixing it?
<whitequark>
I didn't fix anything
<whitequark>
it worked that way from the beginning
<wpwrak>
hmm, but there was some problem with it. one where you said it ended up using the display chord tolerance.
<wpwrak>
anyway, i'll holler if i hit anything ;)
<whitequark>
yes, but that one's not solvable.
<whitequark>
(until we get a more proper file format and/or hierarchical regeneration)
<whitequark>
(probably the former)
_whitelogger has joined #solvespace
L29Ah has quit [Remote host closed the connection]
L29Ah has joined #solvespace
<wpwrak>
neat ! imported a DXF, i get "Too many unknowns in a single group!" ;-)
<whitequark>
yes
<whitequark>
delete some constraints
<wpwrak>
i guess to to there only being H and V but no distance constraints
<whitequark>
actually, hang on
<whitequark>
I'm not sure if that'll help
<wpwrak>
i think it can only make it worse ;-)
<whitequark>
nah
<whitequark>
by "unknowns" it doesn't mean "degrees of freedom"
<whitequark>
it means "variables in the system of equations"
<wpwrak>
ah, i see. it's not THAT compex, though. just a PCB outline, exported by kicad pcbnew
<wpwrak>
it makes solvespace act weird. i tried to "isolate" it by adding another group on top, but it just remains there, editable
<whitequark>
yes
<whitequark>
it's not supposed to be used for that
<whitequark>
you don't want "import"
<whitequark>
you want "linking"
<whitequark>
but we don't have DXF linking
<wpwrak>
any other linking for a format kicad can generate ? i have STEP, gerber, VRML, SVG, ...
<whitequark>
we don't have linking of anything except solvespace files
<wpwrak>
just what i though :(
<whitequark>
so hm
<whitequark>
I would like to see IDF linking
<wpwrak>
any suggestion for getting a 2D drawing into a 3D model ? doesn't have to be very precise
<whitequark>
you've said you'll sigh whichever CLA I want, right? (given that it makes sure the code will be distributed under GPLv3 in perpetuity)
<wpwrak>
maybe i'll have some luck with meshlab ...
<wpwrak>
yup
<whitequark>
so, if you wrote a PR to link IDF files I'll merge it
<wpwrak>
hehe ;-)
<whitequark>
roughly you'd have to add a new kind of group, add an IDF reader, then create surfaces based on whatever's in the file
<whitequark>
probably using the code from extrude groups as an outline
<wpwrak>
okay, dunno if i'll have time for that, though. what i'm after is something actually quite simple: i want to update the drawing on the front page of anelok.com
<wpwrak>
the "OLD VERSION !" is getting a bit embarrassing :)
<wpwrak>
for that, i need a wireframe model of the case (the top part), with the PCB for component locations. that'll go into a bsackground image or such. then i'll trace the things i actually want in xfig.
<whitequark>
oh
<whitequark>
use an image request
<whitequark>
the newly added
<wpwrak>
but for finding the right angle where things don't obscure each other, i need a combined 3d model
<whitequark>
so hm
<wpwrak>
the image can be linked to the 3d model, and rotated, etc., with it ?
<whitequark>
let's see
<whitequark>
yes, the image request can be linked to the 3d model, and rotated with it
<whitequark>
the old "background image" feature did not let that
<whitequark>
wpwrak: anyway if you wanted proper IDF linking then you could try motivating (as in, harassing) me until I do it
<whitequark>
it's a really easy feature, comparatively, with a high payoff
<wpwrak>
heh, you already seem *very* motivated ;-)
<wpwrak>
but lemme see how things go with the image. that looks promising.
<whitequark>
well not especially, you should see my inbox right now
<whitequark>
feels like drinking from a firehose
<wpwrak>
return -ETOOMUCHFANMAIL; // :-)
<whitequark>
if only
L29Ah has quit [Remote host closed the connection]
<wpwrak>
image is great ! would be nice to also have, when selecting the image, links to the points that define it. so that one locate them quickly when in the middle of a complex design.
<whitequark>
right, that's also the part of that issue for circle-center and so
<wpwrak>
yup, circles/arcs could use a bit more love, too. and then there's the small issue that selecting a point doesn't show all the "same' points