<InPhase>
peepsalot: All those "add to milestone" "remove from milestone". :)
<InPhase>
I mean I understand. We get stuff that goes analogously. But it seems like these kernel choices and the implications definitely got away from the project's grasp.
<InPhase>
They estimate the overhead of "exactness" as 25% to 80%.
<InPhase>
"At first sight, this may look like being different in the wrong direction; indeed, naive use of CGAL may result in all the robustness problems indicated above, including crashes at runtime."
<InPhase>
Yes, indeed.
<peepsalot>
grr, i'm getting aggravated again
<peepsalot>
pretty sure CGAL is the most headache inducing library i've ever had to deal with
<dalias>
:-p
gunnbr has joined #openscad
gunnbr__ has joined #openscad
gunnbr has quit [Ping timeout: 252 seconds]
gunnbr__ has quit [Read error: Connection reset by peer]
gunnbr__ has joined #openscad
gunnbr__ has quit [Ping timeout: 260 seconds]
gunnbr__ has joined #openscad
rue_mohr has quit [Remote host closed the connection]
gunnbr__ has quit [Ping timeout: 252 seconds]
gunnbr__ has joined #openscad
ur5us has joined #openscad
<gunnbr__>
status?
<othx>
Gthx.NET version 2.08 2021-08-14: OK; Up for 2 days, 21 hours, 25 minutes, 24 seconds; mood: pretty good.
othx has quit [Remote host closed the connection]
othx has joined #openscad
<gunnbr__>
status?
<othx>
Gthx.NET version 2.22 2021-10-04: OK; Up for 24 seconds; mood: pretty good.
* gunnbr__
cheers!
gunnbr__ has quit [Ping timeout: 252 seconds]
snakedLX is now known as snaked
ur5us has quit [Ping timeout: 245 seconds]
gunnbr has joined #openscad
gunnbr has quit [Read error: Connection reset by peer]
gunnbr has joined #openscad
gunnbr has quit [Ping timeout: 260 seconds]
snaked has quit [Quit: Leaving]
snaked has joined #openscad
gunnbr has joined #openscad
gunnbr has quit [Ping timeout: 260 seconds]
arebil has joined #openscad
Jack21 has joined #openscad
ur5us has joined #openscad
ur5us has quit [Remote host closed the connection]
TheAssassin has quit [Remote host closed the connection]
TheAssassin has joined #openscad
<peepsalot>
teepee: whats the latest situation with CGAL versioning? i remember some discussion about version 5+ moving to header only, which should make packaging easier? can we reliably depend on latest versions?
<peepsalot>
or are we beholden to debian versioning?
<teepee>
ah, difficult question, in theory all the builds are using latest CGAL
<teepee>
uh, all except the windows builds?
<peepsalot>
openscad-nightly shows library info CGAL 5.0.2 for me
<peepsalot>
5.3 is latest
<peepsalot>
teepee: also, without the default cmake build enabling it in some way (ie including as git submodule is only one that comes to mind), then any developer's local builds would not have latest either
<peepsalot>
but all i know is when I've suggested submodules in the past i'm told that overriding a package which they have in repos would make debian very mad at us.
<peepsalot>
i'm not trying to stir anything up, but would appreciate some clarification on the whole submodule situation I guess.
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
arebil has joined #openscad
<teepee>
hmm, i have CGAL version, kernels: 5.2, Cartesian<Gmpq>, Extended_cartesian<Gmpq>, Epeck
<teepee>
which package version is that?
<peepsalot>
yeah, i omitted the kernel info for brevity. OpenSCAD Version: 2021.10.06.nightly (git dc045a7) ... CGAL version, kernels: 5.0.2, Cartesian<Gmpq>, Extended_cartesian<Gmpq>, Epeck
<teepee>
I mean which package? ubuntu 20.04?
arebil has quit [Ping timeout: 265 seconds]
<peepsalot>
yeah
<teepee>
maybe it's not finding the newer cgal dependencies on OBS
<teepee>
ok, it's building, lets see if that works
<teepee>
I don't know if git submodules are a good way to solve this, they seem to be annoying at times, but I'm not completely against using those
<teepee>
I mean we already use it for MCAD
<peepsalot>
yeah, i just figured that one got a pass because its owned by us
<teepee>
my worry is mostly that then there's no wiggle room anymore for platform specific stuff
<InPhase>
peepsalot: Current Debian stable is on 5.2.3, and it from the Ubuntu package pool, it looks like the 22.04 LTS will be on 5.2.3 or later.
<InPhase>
peepsalot: So I think we can count on that as a minimum going forward for the Linux targets.
<InPhase>
We can reasonably advance-target the upcoming releases I think.
<InPhase>
I see no clarity anywhere on when 5.3 would roll out.
<teepee>
strange, got a build fail mail, but it seems to still build fine
<teepee>
huh, cmake says: [ 157s] -- CGAL: 5.00
<teepee>
so it's getting somethings else :(
<peepsalot>
InPhase: 5.3 has been released https://www.cgal.org/2021/07/06/cgal53/ do you mean there was "no clarity" like ahead of time as to a planned release date?
<InPhase>
peepsalot: If it matters when 5.3 rolls out, contacting the maintainer would be reasonable. I don't see any evidence of an IRC presence for the libcgal-dev package maintainer, Joachim Reichel, but his github avatar is a menger sponge, so he must be a decent person. :) His email is reichel@debian.org
<InPhase>
peepsalot: I mean no clarity on when it would make it into debian, ubuntu, and derivatives.
<peepsalot>
ah
noonien has quit [Ping timeout: 252 seconds]
noonien has joined #openscad
<InPhase>
I assume they follow a complex dependency tree on up the line to figure out when they can boost versions and to what.
noonien0 has joined #openscad
Guest6014 has joined #openscad
noonien has quit [Ping timeout: 252 seconds]
noonien0 is now known as noonien
<teepee>
argh, cgal 5.3 needs boost 1.71 :(
<teepee>
hmm, no I think I'm reading that error wrong, it seems to have boost 1.71 but is missing a subcomponent or so
<teepee>
right, tech twitter is strange. lots of retweets for graphscad which does sound interesting but is a gitrepo with 4 commits on a single day in 2017
<teepee>
I do like the idea, but it would be even nicer to be more like IceStudio for HDL so nodes are in the core language
<othx>
peepsalot linked to YouTube video "You can SKIP 3D modeling now" => 1 IRC mentions
<InPhase>
Oh my. These GraphScad screenshots are giving me LabVIEW flashbacks.
<peepsalot>
SPOILER: for a glorified excel spreadsheet...
<teepee>
oh, is that this gcode generator thingy?
<peepsalot>
yeah
<peepsalot>
sad that its only compatible with MS Excel
<teepee>
I've not seen the video as I mostly disagree with his opinions but I've seen some news about the tool and was looking in wonder at the github repo seeing the excel sheet
<InPhase>
peepsalot: "Right after we check out today's sponsor, the people that destroyed freenode."
<teepee>
it's a good addition to the tool box feature wise as some things are probably better to be expressed directly
<teepee>
nah, PIA? they don't have anything to do with that for years
<InPhase>
Does he no longer own it?
<teepee>
I don't think so
<teepee>
"In November 2019, Private Internet Access was acquired by Kape Technologies." no idea who that is
<InPhase>
Well, alright. I'm going to remain on guard about those guys for a while anyway, but I will refrain from besmirching them then if he's actually no longer affiliated.
<peepsalot>
before I watched it, i thought the vid was gonna be about some crazy AI that can model stuff for you
<teepee>
well, being wary of the VPN business in general should be a good thing :)
<peepsalot>
imagine my disappointment :P
<teepee>
yeah, that's an even deeper drop than I had as I just expected some python script which would be fine
* teepee
crosses fingers for snap to build with cgal 5.3...
<InPhase>
peepsalot: Well I endorse more direct gcode generation... But Excel? This is a tool for non-programmers, with a design difficulty process suitable mostly for programmers. Sounds like a worst of both worlds approach.
<peepsalot>
yeah its an odd one
<InPhase>
Overall that video seemed pretty awkward I think. Like trying to teach calculus on an abacus.
<teepee>
aha! [ 196s] -- CGAL: 5.3
<teepee>
cgal cmake failed due to missing boost-program-options and boost-thread
<InPhase>
So will 5.3 become a local build dependency?
<InPhase>
I do not have this yet, but can work on setting this up.
<teepee>
right now it even works with 4.9 I believe
<teepee>
so that would depend on what API we start using going forward
<teepee>
the "blazing fast csg" branch even had a 4.x and 5.x test run
<peepsalot>
teepee: yeah that was one of the things that kept me from doing more testing on it, "fast-csg" feature required 5.1+ and 20.04 is on 5.0
<teepee>
that was the trigger to update the builds to cgal 5.x, I guess it would be worth it
<teepee>
specifically even 5.2.3 as I seem to remember that has some additional fixes for issues found in the fast-csg branch
<InPhase>
RoyK: I don't know if functions for circles and arbitrary functional form traces exist, but that part is so trivial you could finish it in an hour or two including testing and demos.
<teepee>
hmm, that looks way too low level from what I've imagined, is that excel similar?
<InPhase>
That excel appears to be basically that, with one row per thing.
<InPhase>
Except they have rows for "do a circle" and "trace this parametric function".
<InPhase>
Which is trivially addable to the Python thing.
<teepee>
so it only mean you don't have to remember the gcode numbers and have full names?
<InPhase>
Pretty much.
<InPhase>
Also, abstraction so you can use different gcode outputs.
<teepee>
trace parametric function sounds more like what I would want to see
<InPhase>
GCode is actually designed to be very easy to use and to script, it's just ugly.
<InPhase>
If you only want to support one type of output, the interfaces for abstracting it out can be paper thin.
<teepee>
I've generated 2.5D generated from java code :)
<teepee>
the low level assembler part is pretty useful
<teepee>
but for actually using this for designs, some more high level support would be nice
<InPhase>
Right. The spreadsheet did not do that.
<InPhase>
I mean it had loops, but loops are built into that Python approach.
<teepee>
what I had was basically a 2d path that was extruded + a bitmap that was used to imprint the image on the way up