califax has quit [Remote host closed the connection]
califax has joined #openscad
J2275 has joined #openscad
J22 has quit [Ping timeout: 252 seconds]
Guest21 has joined #openscad
Guest21 has left #openscad [#openscad]
LordOfBikes has quit [Ping timeout: 250 seconds]
LordOfBikes has joined #openscad
fling has quit [Remote host closed the connection]
fling has joined #openscad
califax has quit [Ping timeout: 240 seconds]
califax has joined #openscad
Morganite has quit [Quit: Client closed]
fling has quit [Remote host closed the connection]
califax has quit [Ping timeout: 240 seconds]
califax has joined #openscad
GNUmoon has joined #openscad
GNUmoon has quit [Ping timeout: 240 seconds]
GNUmoon has joined #openscad
fling has joined #openscad
noonien has quit [Ping timeout: 256 seconds]
nanoflite has joined #openscad
nanoflite has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
nanoflite has joined #openscad
nanoflite has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
fling has quit [Remote host closed the connection]
fling has joined #openscad
nanoflite has joined #openscad
nanoflite has quit [Ping timeout: 246 seconds]
Alexer has quit [Ping timeout: 250 seconds]
Alexer has joined #openscad
fling has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
fling has joined #openscad
<Scopeuk>
It would be kind of interesting for universities to maybe put up lab time on appropriate courses for mid to advanced programming to do some maintenance work on things that have been run by the university historically. resolving known issues within a system is an industry desirable skill as is bumping into the real world
<Scopeuk>
30 hours of lab for one person isn't going to achieve much but 15 or 20 undergrads doing 30h each probably could
<Scopeuk>
to be honest learning to properly triage bug reports is an incredibly valuable skill
GNUmoon has quit [Remote host closed the connection]
fling has quit [Ping timeout: 240 seconds]
fling has joined #openscad
StudentWeis has joined #openscad
StudentWeis has quit [Client Quit]
GNUmoon has joined #openscad
lastrodamo has joined #openscad
<teepee>
Scopeuk: that's a great idea
<teepee>
If I ever have a bit more time, that could be something to disucss with our local university here
<J2275>
hm should a % object become invisible behind a translucent object? %translate([1,0,-10])cube(5);
<J2275>
color(alpha=.4)cube(10);
<teepee>
that's just the "normal" issue that transparency is order dependent
<teepee>
I think
<teepee>
problem is you can't actually control the order in this specific case
fling has quit [Ping timeout: 240 seconds]
fling has joined #openscad
<J2275>
ah so % moves it from that order
<teepee>
yes, it's taken out of the normal order IIRC
<teepee>
I think those are rendered in a second pass
<teepee>
not 100% sure though
<J2275>
and wouldn't it make sense if ! show objects exclusively without calculating the rest? (like !A();B(); is not equal to A();*B(); )
<teepee>
nope, not possible as there might be some data missing
<teepee>
only the geometry part can be skipped
<teepee>
well, it might be possible to skip *some* stuff
<teepee>
hmm, maybe via some sort of transformation, that might make it possible to get same result for those 2 cases
<teepee>
but in general, I agree it would make sense for those 2 specific cases to do the same :)
<teepee>
not in case the ! is somewhere deeper in the tree
<teepee>
A() !B(); C();
<J2275>
yeah .. i just use ! to concentrate on a single part but the F5 times are too slow so i have to disable all other parts
fling has quit [Ping timeout: 240 seconds]
fling has joined #openscad
StudentWeis has joined #openscad
fling has quit [Ping timeout: 240 seconds]
StudentWeis has quit [Remote host closed the connection]
StudentWeis has joined #openscad
fling has joined #openscad
arebil has quit [Quit: My keyboard has gone to sleep. ZZZzzz…]
StudentWeis has quit [Remote host closed the connection]
StudentWeis has joined #openscad
StudentWeis has quit [Remote host closed the connection]
StudentWeis has joined #openscad
StudentWeis has quit [Client Quit]
califax has quit [Remote host closed the connection]
<teepee>
hmm, not finding that name at least, need to check the whole thing out again
fling has quit [Ping timeout: 240 seconds]
fling has joined #openscad
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
<J2275>
idST_ResourceIDrequiredDefines the unique identifier for this object.
J2275 has quit [Quit: Client closed]
J2275 has joined #openscad
arebil has joined #openscad
<peepsalot>
J2275: looks neat but what does it do?
<J2275>
it is a hinge or a clamp that can be set by turning the grip ring
<J2275>
due to the ratio it is self locking
<peepsalot>
what is the ratio?
<J2275>
20:8 and 19:7
<J2275>
so the between the rings one turn of the grip move the ring 2.5 teeth
<J2275>
or 45°
<peepsalot>
uh, so 8:1 then (360:45)?
<J2275>
yes
<peepsalot>
how do those two ratios combine to make that?
<J2275>
an earlier version had a grip at the gear itself .. which was a bit annoying
<peepsalot>
is it actually ~6.79:1
<peepsalot>
doesn't seem like crazy high to the point where I would expect it can't be backdriven
<J2275>
i turn schould be 47,368....°
<J2275>
or 1: 7,6
<J2275>
it is dopenden if you hold the bigger or smaller ring when turning
<J2275>
i think you are right with 6.7857..
<peepsalot>
i just multiplied the two ratios you gave
<peepsalot>
20/8 * 19/7
<J2275>
that is simpler then what i did Ü
<J2275>
you can generate higher ratios which i think are not practical as rotating is getting too slow
<J2275>
hm maybe you need to account for the rotation of the gears to as the second ring is driven reverse
fling has quit [Ping timeout: 240 seconds]
<J2275>
peepsalot just tested it and rotating the center 7× is opening the clamp ~ 180°
<J2275>
not sure how to calculate that 12:1 ratio
<J2275>
or 14:1 holding the other ring
fling has joined #openscad
<J2275>
11,66666
<J2275>
2.5×7/19-1
<J2275>
19/(2.5*7) - 1
aiyion has quit [Ping timeout: 240 seconds]
<J2275>
damn so close to 1:12 .. so it will not be a clock Ü
aiyion has joined #openscad
<J2275>
with a 14 and 13 ring gear and a 7/6 planet i get 1:12 ratio
<J2275>
27 and 9 is probably easier to build (or 40/39 and 10/9)
<joseph_>
I was rebuilding OpenSCAD after a minor UI change and got a strange segfault from the compiler. I just ran make again with no changes and it worked the second time. Should I post the output in a pastebin?
<dTal>
you should check your ram :p
fling has quit [Quit: ZNC 1.8.2+deb2+b1 - https://znc.in]
fling has joined #openscad
<InPhase>
joseph_: If the compiler ever segfaults, that's never the fault of the source code being compiled. It's a compiler defect.
<InPhase>
A properly working compiler can be fed anything up through randomly generated text and it will produce either a compiled result or a sequence of error messages.
<joseph_>
InPhase: Makes sense. Is it still worth keeping a record of the error?
<InPhase>
joseph_: Only if you are going to try to report it as an issue on the compiler. But that would really only help them if you can reproduce it.
<InPhase>
Sometimes it's hard to reproduce an issue that happens on a rebuild though, unless you know the exact steps to get to the same rebuild state.
<InPhase>
I don't remember the last time I got a compiler segfault, but I do get anomalous output sometimes on rebuilds. The first reflex step on that should always be to attempt a clean build.
<InPhase>
(When your core goal is just to get things working.)
<joseph_>
InPhase: Ok. Probably not worth the effort to report to GCC unless it happens again
<joseph_>
Also my manual testing indicates I have a fix for #3706, should I make a PR? Also any conventions I should follow e.g. linking to the original issue
<InPhase>
Yeah, if your text says "fixes #3706" as part of a sentence, then github will detect that text and mark it as a fix for this.
<InPhase>
Then once you submit it, we wait for the automated building and testing to complete, which can take an hour or so, and then you check back at that to see if everything went smoothly.
<InPhase>
This will appear toward the bottom of the issue.
<InPhase>
Include any other text in your description that's necessary to document subtleties in the PR, explain the specific changes in the behavior in a summary, and note any new issues or concerns it might introduce if there are any.
<InPhase>
I've made PRs where the issue text was as simple as "Fixes #somenum", with adequate detail in the title. But sometimes you need a lot more. Those are just the lists of things to think about for needing to be in there. On a project of this size, the PR text becomes a living record for people to look back to whenever things go wrong later with a particular change. Like, "This attempted to change X,
<InPhase>
but Y was changed too, so that seems to have been an unintentional bug."
<myosotis>
I would hide my head in the sand and pretend like I didn't see anything
<myosotis>
until it happened more than one time in my life
<Scopeuk>
Also for "this appear to use a needlessly complicated system rather than the trivial obvious solution" allowing you to check if it's there for am very specific reason, ideally code comments would cover that though
<joseph_>
How important is it that commit history I've made on the branch to be merged are clean and logical? For example I made a commit, reverted it, but then rebased on master, so the commit ID in the description is no longer valid
<teepee>
joseph_: did it say something with "pipe" ?
<teepee>
then... what dTal said :)
<teepee>
there's a couple of files that need quite some ram to compile, if you run parallel build and hit multiple such files at the same time, it can crash the compiler with out of memory
<teepee>
specifically any combination of the cgal-*.cc files
<teepee>
that said, -j12 seems quite optimistic. how much ram is in that VM?
<teepee>
or docker container?
<joseph_>
It's a VM with 24GB ram
<joseph_>
The first time I ever tried building, my system did run out of memory. It was my own fault for leaving a large file in a ramdisk. But in that instance it wasn't a compiler segfault, what happened was that the host OOM killer just shut down the entire VM
<InPhase>
Well it bailed out inside the stdlib too there. :)
<teepee>
with gcc, I'd say it's maybe a bit too much with -j12
<teepee>
it might work ok with clang
<InPhase>
Match -j to cores if you have large buckets of RAM and good thermals. Otherwise it can go better to scale down.
<teepee>
on 16gb and gcc, I normally keep it at a conservative 4
<teepee>
with lots of stuff running, so there's not actually 16gb available :)
<InPhase>
I do -j8 with 8 cores and 64GB of RAM, and never hit an issue. But my last laptop had to stay much lower...
<InPhase>
joseph_: You can tweak them to your system's needs, and thus not need to remember your preferred settings after that.
<joseph_>
Thanks. The host machine is a desktop with 12c/24t and 48GB. Once I've gotten Docker working and no longer need the VM, maybe I will try higher parallelization. But for now I'll tone it down
<InPhase>
joseph_: I personally prefer to skip the "make install" and then instead copy the executable out of build/ into the top level, and I run each repository's openscad executable from there. This is more useful when you're juggling comparisons between a few branches at once.
<teepee>
why copy?
<InPhase>
Because then I always have the last one that successfully built, even if a clean build failed.
<teepee>
ah, ok, scary :)
qeed_ has joined #openscad
<InPhase>
Well, for example my masterscadvim script is setup to use the latest openscad in ~/src/openscad-master/ :)
<InPhase>
So if teepee goes and breaks the master branch, I keep on using the last good one.
<teepee>
master never breaks :P
<InPhase>
"never"
<teepee>
it may do some strange things in rare occations but that's for testing purposes ;-)
<teepee>
just like my debian testing which breaks every 2 years to not being able to login
<InPhase>
But yeah, it also gives the same benefit for the more unreliable branches, like new tricky features.
qeed has quit [Ping timeout: 246 seconds]
<teepee>
which usually marks upgrades of gnome-shell :)
<InPhase>
I did in fact have the master build break once in the last year, with that CMakeLists Qt 5 finding update I PRd. Was not too hard of a fix, but you know, just in case.
<teepee>
that's a build-from-source break, so local issue :P
<InPhase>
Yes, the kind that matters to me. :)
<teepee>
yeah, and that's good and helpful
<InPhase>
Also, free bonus, if I am typing it manually I don't have to type the build/. Like, ~/src/openscad-pr4151/openscad instead of ~/src/openscad-pr4151/build/openscad Better to let the computer handle that extra step with a copy.
<teepee>
installing openscad-nightly takes care of that as it just can't update if the builds fail on OBS
<teepee>
we actually broke the Snap build a while ago, and for some strange reason they don't send mail notifications for that
<joseph_>
It looks like the pipeline for my pull request passed. Let me know if my comment on it makes sense. I've identified another way to solve the issue, so let me know if you think I should take that alternate approach instead
<teepee>
I'll try to have a look after groceries shopping
<teepee>
although it's more likely I just fall asleep ;-)
<InPhase>
The code change looks pretty trivial. I'm building to attempt a usage test.
<InPhase>
joseph_: Ah, also the auto build and test didn't run, as this was your first PR. I clicked the "Approve" button. Now they should run for you automatically.
<InPhase>
If you check again you will hopefully see them in progress.
<teepee>
I think the approve only goes away after the first merge
<InPhase>
joseph_: Looks okay. It does all the correct things across the range of behaviors I'm aware of.
<InPhase>
We can merge as soon as the auto-tests finish successfully. (They probably will do fine on this. There should be no opportunities to break them.)
<teepee>
yep
<teepee>
sometimes they fail due to network issues, which can be fixed by just restarting the builds :)
<teepee>
or more annoyingly when homebrew updates at random times
<teepee>
like with ghostscript a couple of days ago... not the fault of ghostscript though
<teepee>
almost the same fix I needed today when doing some production installation
<teepee>
the earlier rollout managed to get out a broken version with a "}" missing, cool stuff
ShakespeareFan00 has joined #openscad
<ShakespeareFan00>
Hi
<ShakespeareFan00>
Apologies in advance is this is offtopic but does anyone here know of an OepnSCAD type tool for making card models?
<InPhase>
What sort of card?
<ShakespeareFan00>
paper.
<ShakespeareFan00>
As in I'd like to be able to take a low poly STL model and be able to print it out
<teepee>
as in unfolding?
<ShakespeareFan00>
That and simplistic uv mapping against vector based artwork.
<peepsalot>
STL is 3D format, paper is 2D, kind of hard to make sense of what you are asking
<ShakespeareFan00>
I want to take a very low vertex STL and unfold it.
<peepsalot>
STL doesn't have uv mapping
<ShakespeareFan00>
bother.
<InPhase>
Someone in here was working on unfolded models at one point with some sort of library.
<teepee>
well, you could uv-unmap an STL I suppose
<ShakespeareFan00>
Okay let me step a back a bit and exlpain the use case...
<teepee>
but not in OpenSCAD
<InPhase>
That was maybe 2 years ago?
<peepsalot>
only open source vector graphics app i know is inkscape
<ShakespeareFan00>
Years ago on an old computer called the Acorn A3020, there was a progrma called !Tabs
<ShakespeareFan00>
This let you build up a 3D object using constructive sold geometry
<ShakespeareFan00>
And then print it out for assembly on 'card-stock' or heavy printer paper.
<InPhase>
Found it in the logs. One year ago. That person who was doing it last April was named ShakespeareFan00.
<ShakespeareFan00>
Oh...
<dTal>
that's an astonishing coincidence
<ShakespeareFan00>
That was me :(
<ShakespeareFan00>
And i didn't get much further
<ShakespeareFan00>
Apologies for wasting your time :(
<InPhase>
At least you have a very consistent April hobby. :)
<InPhase>
I can give you back the links you found last time if you lost them.
<ShakespeareFan00>
As long as you don't think it was an April 1st troll :(
<ShakespeareFan00>
Inphase: Probably better you compiled them down into something on a help page or on a wiki page..
<ShakespeareFan00>
I can't be the only person interested in this
<ShakespeareFan00>
It's a shame that I have lovely ideas but can't ever dcommunicate them well enough to inspire someone to actually the code to do implement them...
<ShakespeareFan00>
Since last April - Inkscape got at least 2 unfolding extensions...
<ShakespeareFan00>
So unfolding the STL isn't as tricky as it was previously
<ShakespeareFan00>
Of course the other idea I had in mind was to change what i wanted to do...
<InPhase>
I guess that was more links than I thought it was going to be. :)
<ShakespeareFan00>
teepeee: Would anyone use it though?
<ShakespeareFan00>
And personally 'unwrap' would make more sense...
<ShakespeareFan00>
teepee: And if you have an unfold, suely you'd also want to have a 'fold' command as well?
<ShakespeareFan00>
To fold a nominally flat panel along a defined axes.
<InPhase>
It would have limited utility probably, as it would be hard to use the output of such a routine as the input in anything else. So it would mostly just be a thing to call before output.
<InPhase>
One can just as easily use an external unwrapping tool at that point.
<ShakespeareFan00>
Inphase that was kind fo what i was thinking...
<ShakespeareFan00>
Are people here familiar with the sort of components like brackets that are bent sheet metal?
<ShakespeareFan00>
(You've set me off an a tangent here)
<InPhase>
It's hard to not be familiar with bent sheet metal brackets. :)
<ShakespeareFan00>
Assuming someone wrote a library, you could have a fold() command. that folded a part along a given axis
<ShakespeareFan00>
Not that OpenSCAD is ever going to be a CAD tool for that sort of component.
<InPhase>
That's an external tool to unwrap stl files into laser cut inputs.
<ShakespeareFan00>
Hmm... Do we have an SCAD offtopic...
<ShakespeareFan00>
?
<ShakespeareFan00>
I had some really big ideas.. ;)
<ShakespeareFan00>
teepee: I started looking into paper models as a way to have a means of generating model buildings...
<ShakespeareFan00>
but this got me thinking as to whether given the availability of 3D printers, I'd be better off having 3d printable parts and sticker artwork
teepee has quit [Remote host closed the connection]
<ShakespeareFan00>
INphase: Thanks.... You've suggested something to me :)
teepee has joined #openscad
<ShakespeareFan00>
many many years ago, there was a model railway brand in the UK called Graham farish
<ShakespeareFan00>
With some of their train sets they supplied model buildings...
<ShakespeareFan00>
in N scale.
<ShakespeareFan00>
These were not card models, but plastic building blocks in the right shape and sticker artwork for the external surfaces
<ShakespeareFan00>
Creating scaled blocks is meat and potatoes work for OpenScad
<ShakespeareFan00>
And (for some textured finishes)
<Scopeuk>
I have unfolded one scad model to make a paper version before I had my 3d printer, that was done using rotate, project and then translate to position each surface adjacent, it wouldn't work for high complexity though
<ShakespeareFan00>
Some specfic detail could then be added with stickers or decals like you would with other forms of plastic model kit... hmmm
<ShakespeareFan00>
I suppose my real problem is how to get a sensible UV map I can use as a template for 'decals' or stickers therefore
<Scopeuk>
I suppose for a model village like setup you could define some primitive modules that had a manually generate unfold and then positioning logic, such that you have building or building map which take the same parameters, one to give a representation of end result and one to give the intermediate product
<ShakespeareFan00>
True...
<ShakespeareFan00>
I'm finding the 3d components and 'decals' approach might be better than trying to a full unfold..
<Scopeuk>
Hmm I wonder if blender has anything to do unwrapping from stl
<ShakespeareFan00>
Annoyingly I can't find any picture online of the Graham Farish type buildings kits
<ShakespeareFan00>
Yes...
<ShakespeareFan00>
Although I'd probbaly be aimign for UK style designs, vs US or EU ones ;)
noonien has joined #openscad
<ShakespeareFan00>
I'll also have another look at Freecad
<ShakespeareFan00>
:)
<InPhase>
ShakespeareFan00: I see the computers that old Designer Castles thing was built for pretty much only ever sold in the UK. That probably explains the sparse interest in reviving it. Just not that many people ran into it in their youth in the narrow window when that existed.
RichardPotthoff has quit [Ping timeout: 246 seconds]
ShakespeareFan00 has quit [Quit: Leaving]
snaked has joined #openscad
splud has quit [Ping timeout: 246 seconds]
splud has joined #openscad
<joseph_>
InPhase: Is my PR ready to merge?
<teepee>
yep, I think it is
epony has quit [Ping timeout: 260 seconds]
fling has quit [Remote host closed the connection]
aiyion has quit [Remote host closed the connection]
fling has joined #openscad
aiyion has joined #openscad
<joseph_>
teepee: I don't see a merge button on my end, even though GitHub now says approval not required
qeed_ has quit [Quit: qeed_]
<teepee>
joseph_: I don't think it's possible to enable that, merge still has to be clicked by someone with write access to the repo regardless of approval
<teepee>
that said, I would not claim to understand all the settings github allows :)