teepee changed the topic of #openscad to: OpenSCAD - The Programmers Solid 3D CAD Modeller | This channel is logged! | 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
mtm has quit [Ping timeout: 252 seconds]
mtm has joined #openscad
RoyK has quit [Ping timeout: 246 seconds]
mmu_man has quit [Ping timeout: 252 seconds]
RoyK has joined #openscad
aiyion1 has quit [Ping timeout: 264 seconds]
aiyion1 has joined #openscad
mmu_man has joined #openscad
<gbruno> [github] kintel pushed 1 modifications (macOS build fix) https://github.com/openscad/openscad/commit/2c5ed91655a1ec6fdf181ea99978df0d1288c0aa
<gbruno> [github] kintel synchronize pull request #5563 (Add parameter list for local application start. Unify dark mode switch.) https://github.com/openscad/openscad/pull/5563
<gbruno> [github] t-paul pushed 2 modifications (Move templated function to header file.) https://github.com/openscad/openscad/commit/5d72d829cffe46a119eb5faf37e6aafec2588b9f
<gbruno> [github] t-paul synchronize pull request #5563 (Add parameter list for local application start. Unify dark mode switch.) https://github.com/openscad/openscad/pull/5563
snaked has joined #openscad
linext has joined #openscad
mmu_man has quit [Ping timeout: 260 seconds]
J24k11 has joined #openscad
linext has quit [Quit: Client closed]
<gbruno> [github] t-paul pushed 123 additions 7 modifications 121 removals (Move icons into light/dark theme.) https://github.com/openscad/openscad/commit/78db9ea4c35b04ac977570afb7abffe8d8f6c1e5
<gbruno> [github] t-paul synchronize pull request #5563 (Add parameter list for local application start. Unify dark mode switch.) https://github.com/openscad/openscad/pull/5563
<teepee> This light/dark mode stuff is annoying :/
J24k32 has quit [Ping timeout: 240 seconds]
<teepee> kintel: could you have a look at the #5563 change, if that works on macOS? I've separated the normal and "-white" icons into distinct generated qrc files and declared them as icon theme
<teepee> with that a simple QIcon::setThemeName("svg-default") or "svg-default-dark" should switch all icons over without that dance we do in many places manually switching resources
<teepee> but I don't want to update all the icons before it's certain it works on all platforms :-)
<gbruno> [github] t-paul pushed 2 additions 1 modifications (Ignore only specific *.qrc files.) https://github.com/openscad/openscad/commit/50cbfbb9e70200df6d15bbdab013a428da56aacd
<gbruno> [github] t-paul synchronize pull request #5563 (Add parameter list for local application start. Unify dark mode switch.) https://github.com/openscad/openscad/pull/5563
<teepee> specifically the toolbar looks https://imgur.com/EvA4OC4 for me, export still broken but the main icons like file-new / file-open are ok
lastrodamo has joined #openscad
kintel has joined #openscad
<gbruno> [github] jordanbrown0 synchronize pull request #5562 (Consistent quoting of variable names (via central function). Fixes #2833.) https://github.com/openscad/openscad/pull/5562
<kintel> teepee This is mine: https://imgur.com/a/58jkHno
<teepee> cool, so that looks like it works
<teepee> annoyingly i can't find a way to make the icons appear in designer
<teepee> but keeping the code cleaner seems more useful and maybe I'll find it eventually
<kintel> No ideas here; I've lost touch with Qt : /
<kintel> Can anyone reliably reproduce the latest (stand-alone but large) repro case here? https://github.com/openscad/openscad/issues/5554
<teepee> whew, thanks for the copy button github :)
<teepee> I'm getting the spiky one, that's the broken case?
<teepee> hmm, we don't show the clipper version. let me check what is packaged
<kintel> yup, the spike is the bug. Not getting spike here..
<kintel> I can look into adding clipper version to info
<teepee> is that header only? I don't see a dependency to polyclipping2
<teepee> _service:recompress:tar_scm:Clipper2-20240514T220151.gitff37866.debian.tar.gz
<kintel> no, but it could be a static lib, depending on platform
<teepee> that's what? 8 month
<teepee> lets retrigger that
<kintel> btw:
* kintel option(USE_BUILTIN_CLIPPER2 "Use Clipper2 from submodule" ON)
<teepee> oh, so it's not using the package anymore?
<kintel> It's a bit unreliable to display version for submodules, as libraries sometimes just keep the last released version in the sources
<kintel> We were never using any package for Clipper2.
<teepee> argh, they changed the whole OBS web gui
<teepee> OBS did
<teepee> for the debian builds
<kintel> Right, for Manifold?
<kintel> We probably need to set that up for OpenSCAD too.
<teepee> yes, I think I introduced it for manifold as OBS has no network access, so cmake fetch is simply no option
<teepee> [ 44s] -- Clipper2 submodule
<teepee> yep, the nightly build seems to use the submodule
<kintel> yeah, best would be to align Manifold's and OpenSCAD's dependency
<teepee> while packaging is annoying, submodules are 10 times more so
<kintel> Actually, I think they should be aligned. OpenSCAD will provide its submodule to Manifold
<kintel> Same goes for manifold itself of course
<teepee> so clipper versions should be same for the recent builds regardless of platform
<kintel> for the purposes of the initial discussion, yes
<gbruno> [github] kintel pushed 1 modifications (Display Clipper2 version as part of library info) https://github.com/openscad/openscad/commit/0155be90569a514d60210e107ef1b23e6a5dbbd3
<gbruno> [github] kintel opened pull request #5564 (Display Clipper2 version as part of library info) https://github.com/openscad/openscad/pull/5564
<teepee> there's tiny edge which gets out of control due to offset with delta = 50
<teepee> added second image with the full delta
<kintel> hm, doesn't immediately strike me as a precision issue. I'll fire up my Intel box and play around with it
<gbruno> [github] jordanbrown0 synchronize pull request #5553 (Start rotate_extrude on -X only for no angle specified.) https://github.com/openscad/openscad/pull/5553
<teepee> it looks a bit like a bug in delta with negative value
<kintel> Btw. for fun, try to up the precision a bit
<kintel> constexpr int DEFAULT_PRECISION = 8;
<kintel> ^ This means minimum 10^8 absolute fixed point precision
<kintel> Clipper2 uses int64_t, so we should have some room
<kintel> Right now it's aligned with Manifold's usage of Clipper2, mostly to avoid surprises when converting between the two
<teepee> with 9 it seems to be gone
<teepee> no spikes with 9 for this specific shape
<kintel> I wonder if this would also fail for a pure Manifold test, prompting a precision upgrade there as well
<teepee> it's very random, spike sometimes going up sometimes going down depending on the delta - value
<kintel> When debugging such examples, I really want a "flatten tree to a polyhedron()" function
<kintel> Heh, managed to reproduce and got a spike extending to -10^6
<gbruno> [github] jordanbrown0 synchronize pull request #5553 (Start rotate_extrude on -X only for no angle specified.) https://github.com/openscad/openscad/pull/5553
<kintel> paste?
<othx> paste is https://bpa.st for .scad files and other text like long error reports, https://pasteboard.co/ or https://imgur.com/ for images
<kintel> With this minimal repro, I get a large spike, also on macOS: https://bpa.st/L5QA
<teepee> lol, large! that's poking the next galaxy
<gbruno> [github] kintel pushed 1 modifications (Display Clipper2 version as part of library info (#5564)) https://github.com/openscad/openscad/commit/e23f45f84bbf7ae0d070a426e6fc72ad7fb70f33
<gbruno> [github] kintel closed pull request #5564 (Display Clipper2 version as part of library info) https://github.com/openscad/openscad/pull/5564
JakeSays has quit [Ping timeout: 265 seconds]
teepee_ has joined #openscad
teepee has quit [Ping timeout: 264 seconds]
teepee_ is now known as teepee
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
rawgreaze has quit [Ping timeout: 252 seconds]
<gbruno> [github] t-paul converted_to_draft pull request #5563 (Add parameter list for local application start. Unify dark mode switch.) https://github.com/openscad/openscad/pull/5563
<InPhase> kintel: Nice. I was trying to do a minimal reproduction, but that starting file was a real mess for a bug report. :) And it was clearly a floating point edge issue.
<InPhase> Your reproduction makes clear it's an actual regression though, and not just a floating point outcome difference.
<InPhase> Although presumably floating point proximity triggered somewhere inside.
JakeSays has joined #openscad
guerd has joined #openscad
guso78k has joined #openscad
<guso78k> when does it make sense to add parameters to the local slicer invocation ? can't it be encoded in the executable name ?
hyperair has quit [Ping timeout: 252 seconds]
<gbruno> [github] UBaer21 opened issue #5565 (Delta offset creates artifacts) https://github.com/openscad/openscad/issues/5565
guerd has quit [Read error: Connection reset by peer]
<J24k11> kintel I assume while offset need to calculate the angle - for very shallow angles maybe precision is not that high for that calculation and moves the point far away as Tangens changes a lot for small changes near 90.. so if 89.99 is needed but 89.999 is used your point will spike far outside
<J24k11> and in the past multiple offsets can create small jagged edges  - which are seeds for spikes due to the small angles.
<J24k11> maybe some rounding or raster  so  two points that are just 10⁻¹⁵ apart will become one
<J24k11> offset(delta=0.5)polygon([[+0.1,0],[0,0],[0,10]]);
<J24k11> the spike itself is not a bug it is what a delta offset does
knielsen has quit [Quit: leaving]
<J24k11> and normally a following minus offset should remove the spike except some floating math causes them to be different like here offset(-0.4999)offset(delta=0.5)polygon([[+0.00003,0],[0,0],[0,10]]);
<J24k11> if you compare these two offset(-0.4999)offset(delta=0.5)polygon([[+0.00003,0],[0,0],[0,15]]);translate([1,0])offset(-0.4999)offset(delta=0.5)polygon([[+0.00003,0],[0,0],[0,16]]);
<J24k11> the second is shorter .. but should be longer
<J24k11> for(i=[0:50])translate([i,0])offset(delta=0.1)polygon([[+0.00003,0],[0,0],[0,i]]); // this should look different
<J24k11> under a certain angle the spike is much longer as it should be
knielsen has joined #openscad
mmu_man has joined #openscad
mtm has quit [Ping timeout: 260 seconds]
mtm has joined #openscad
dingodoppelt has quit [Remote host closed the connection]
dingodoppelt has joined #openscad
<guso78k> J25k11, I suspect that Clipper2 combines 2 points as one accidently and thus yields wrong result. lowering epsilon values might help
<gbruno> [github] choller opened issue #108 (Issue with DXF import + linear_extrude) https://github.com/openscad/openscad-playground/issues/108
rawgreaze has joined #openscad
<J24k11> guso78k strange is how this changes if the angle gets smaller - you can see there is an wrong behavior for i < 15
Guest10 has joined #openscad
hyperair has joined #openscad
<teepee> guso78k: what did you mean withe the parameters earlier?
Guest10 has quit [Ping timeout: 240 seconds]
little_blossom has quit [Quit: little_blossom]
little_blossom has joined #openscad
<meshugga> I'm going insane trying to create a 2d plan
<meshugga> as soon as i try to add text, everything breaks. I've tried to use a projection, only works without text. then i put in a day of work to re-create everything with 2d polygons... text (with or without projection) doesn't render.
<meshugga> does anyone have a pdf/svg-able example that uses polygons and overlaid text successfully?
<teepee> the overlay parts sounds like a problem
<teepee> OpenSCAD is a "solid" modeler, meaning it enforces closed polygons always, it's not a drawing tool with multiple 2d layers
<meshugga> ignore "overlaid", i need visible text on 2d polygons or a projection
<meshugga> when exported to pdf/svg
<teepee> same thing. separate at the side is ok, differenced (cut-out) is fine. just putting text "into" a polygon is not possible
<teepee> it will union both, so text will merge into the polygon
<meshugga> gaaaaaaaaaaaaaah another project i've put hundreds of hours into which fails at the last moment with something i thought would obviously be supported
<meshugga> jesus christ
<teepee> maybe someone can suggest something if you post an example
<teepee> it may may be possible with some additional step
<meshugga> the example is really just polygon() and text()
<teepee> plus for multi-part/multi-color topics there probably also changes will be coming
<meshugga> I've put a massive amount of work into calculating cut angles and templates for wooden lamps after several (configurable) bezier curves
<meshugga> and I want to be able to print the results (which are there mathematically) onto paper so i can glue it to wood and cut
<teepee> one example I can show is the icons used in OpenSCAD, it's one scad file creating all the icons
<meshugga> basically, I want to have polygons with text
<teepee> but there is a small script that will generate them as separate icons for use in the app
<teepee> what you seem to need is the opposite, generating the "layers" as separate export and unify as 2nd step
<meshugga> apparently i need to projection()difference() {linear_extrude()polygon();linear_extrude()text();}
<gbruno> [github] jordanbrown0 synchronize pull request #5562 (Consistent quoting of variable names (via central function). Fixes #2833.) https://github.com/openscad/openscad/pull/5562
<teepee> you can't mix 2d and 3d but this should not be needed
<J24k11> meshugga  export the polygons .. then the text .. combine the two files in inkscape on different layer
<teepee> the difference of polygon (= 2d) and text (= 2d) should be fine
<teepee> do you have a failing example that's not like 1000 lines?
<teepee> paste?
<othx> paste is https://bpa.st for .scad files and other text like long error reports, https://pasteboard.co/ or https://imgur.com/ for images
<J24k11> if you have text and polygons as separate modules - a difference should be easy
<teepee> if it's more than 2 lines, use a paste service
<meshugga> J24k11: you can not be serious
<J24k11> ?  you don't need extrusion and projection .. the difference works in 2D
<meshugga> J24k11: i mean exporting and using inkscape
<teepee> lets first find if there's something strange going on. as said already that last example should not need the extrudes and projection
<meshugga> teepee: you're right, difference between clean 2d polygons and text works too
<J24k11> you can use any software that combine svg/dxf  - openSCAD is just not supporting layer for export
<J24k11> But you can use "lazy union" then overlapping works in 2D
<teepee> ah, that works for 2d too? hmm I never tried that
<teepee> it's a hack and needs a proper solution, but for now it can be useful
<teepee> I need to try how the svg looks like
<J24k11> teepee here https://imgur.com/a/Mvl3hBv
<J24k11> (added inkscape)
<teepee> well, it's obviously not "layers" as SVG does not even *have* layers, but they show up separately
<J24k11> just overlapping .. you may need to pay attention to the order -- having the text on top
ecraven has quit [Quit: quit]
ecraven has joined #openscad
<J24k11> hm  a layer() module like color() could be used to save polygons on a certain layer
<teepee> I'm still for separate annotations ;-)
<teepee> there's way too many attributes to grab a fixed module name for everything
<J24k11> is there a reason we are not exporting colors in svg?
<teepee> yes, we don't have colors yet
<J24k11> some one said that for 3D a while ago - Ü
<teepee> the new color data structure is not used for 2d I think
<teepee> yes, we have colors for 3d
<teepee> I'm not sure if it's a different data structure or if it's just not implemented yet
<teepee> actually I'm not sure clipper can do that, if you do union it will just merge things
<J24k11> at the moment the polygon also have stroke =1 (or so) which could be removed if you have colors
<teepee> manifold allows keeping the colored faces from the input
<teepee> yes, and/or have an option dialog
<J24k11> or this is defined via annotations
mmu_man has quit [Ping timeout: 260 seconds]
<gbruno> [github] kintel pushed 35 modifications (Consistent quoting of variable names (via central function). Fixes #2833. (#5562) * Consistent quoting of variable names (via central function). Fixes #2833. * Add TODO note for duplicate-assignment messages.) https://github.com/openscad/openscad/commit/bf9188fb1a51ef58bf38351823f84a1b1829f396
<gbruno> [github] kintel closed pull request #5562 (Consistent quoting of variable names (via central function). Fixes #2833.) https://github.com/openscad/openscad/pull/5562
<gbruno> [github] kintel closed issue #2833 (Format warnings - delimit variable names etc.) https://github.com/openscad/openscad/issues/2833
mmu_man has joined #openscad
guso78k has quit [Quit: Client closed]
teepee_ has joined #openscad
teepee has quit [Ping timeout: 264 seconds]
teepee_ is now known as teepee
<teepee> ok, I found a way of showing the icons in the designer, although I don't know if that works on other platforms than Linux
<teepee> but I guess better than nothing
<nihil> too modest, linux is better than everything
<nihil> far from perfect, but with the competition being such shit...
<nihil> (had to use windows for work recently, still ranting)
<teepee> haha, I agree, but I don't mind if other people don't :)
<teepee> and for the record, the icon theming in Qt is just a mess
<nihil> qt has always been a bit fucky
<teepee> unfortunately the options are pretty limited
<nihil> seems to have gotten better in recent years
<nihil> at least a bit more predictable at any rate
<nihil> but I also avoid writing gui things like the plague
<teepee> their new found focus on mobile is a big worry though
<teepee> hmm, so how would I call the icon theme of openscad
<nihil> openscons
<nihil> lots of techy looking glyphs with no fillets or chamfers
<teepee> what about Chokusen? japanese for "straight line" which seems ok as openscad does lines much better than curves :-)
<teepee> any japanese speakers around? if that maybe has some other meanings
<nihil> as a long time lover of wabi sabi, I approve
<nihil> I learned some japanese, but not really conversational or suitable for determining nuance
<nihil> same with many other languages
<teepee> cbmuser you around for some japanese question?
<teepee> argh, it's still picking up the wrong icons, it does not fun though. like april fools version... https://imgur.com/gm1ON17
<nihil> eh, that's just scad on hard mode
L29Ah has left #openscad [#openscad]
<teepee> like those keyboards with no text on the keys :)
<teepee> aha! icons!
<teepee> now lets see if the dark mode switch works
<teepee> yay!
L29Ah has joined #openscad
<teepee> good, no more messing around with dark mode for the icons... https://imgur.com/a/XMyABM8
<nihil> looks good
<nihil> except for that weird panel on the left with the code in it
<nihil> openscad has a built-in editor?
<nihil> :)
<nihil> honestly, being able to use vim is one of my favorite things about openscad
kintel has joined #openscad
<InPhase> nihil: It is a critical feature. :) I will always support it strongly.
<InPhase> nihil: However, the built-in editor has gained some nice features actually. Not that there's anything wrong with that, if that's the way you are.
<InPhase> teepee: lol. I like the gui version where every icon is the OpenSCAD logo. You should try to arrange that on purpose for the 2025-04-01 nightly, and see if anybody notices.
<nihil> lol
<teepee> :)
JordanBrown has joined #openscad
<JordanBrown> KINTEL?
<JordanBrown> kintel?
<kintel> JordanBrown I'm semi-here
kintel has quit [Read error: Connection reset by peer]
<JordanBrown> (sorry, phone call)
<JordanBrown> Sorry, SIGMOM is a high-priority interrupt.
<JordanBrown> kintel (wake up again :-)
<JordanBrown> I ran ctest -C All, and got 44 failures. I don't routinely run -C All - I don't know if I ever have - and so I don't know how many failures to expect.
<gbruno> [github] jordanbrown0 opened issue #5566 (Consistent and unambiguous quoting of file names in messages) https://github.com/openscad/openscad/issues/5566
<InPhase> JordanBrown: Not sure if anyone is even tracking failures in the full set.
<InPhase> JordanBrown: Some might be changes that aren't "failures" but just were never paid attention to updating.
kintel has joined #openscad
<kintel> JordanBrown Ideally, none should fail, but we don't track that automatically
<gbruno> [github] jordanbrown0 opened issue #5567 (Consistent and unambiguous presentation of module and function names in messages) https://github.com/openscad/openscad/issues/5567
<JordanBrown> Where can I find the set that the CI runs?
kintel has quit [Client Quit]
<gbruno> [github] jordanbrown0 opened issue #5568 ("UI-ERROR" prefix seems ... unclean) https://github.com/openscad/openscad/issues/5568
kintel has joined #openscad
<kintel> JordanBrown It's all in the .github folder
<kintel> I believe we run the default set for PRs, and examples only for master pushes
<JordanBrown> I'll look. For some reason I'd accumulated a list of exclusions in my wrapper script.
<kintel> ..so perhaps just ignore All for the purposes of your PR
<JordanBrown> #exclude="-E 'opencsgtest_issue1165|opencsgtest_issue1215|throwntogethertest_issue1089|throwntogethertest_issue1215|openscad-nonascii|import_3mf|3mfpngtest|3mfexport'"
lastrodamo has quit [Quit: Leaving]
<teepee> I think I could get rid of all the manual "is-dark-mode" except the single one flipping the icon theme
<JordanBrown> Ooh, cool, SIGSEGV.
<kintel> JordanBrown Where is that exclusion list from?
<JordanBrown> I haven't the faintest idea.
<JordanBrown> I know that at one point I had problems with building with 3MF support; that certainly could have been part of it.
<JordanBrown> SIGSEGV: https://bpa.st/O7VQ
<JordanBrown> But loading it into that same build and F6 works fine.
<JordanBrown> Clearly not related to my rotate_extrude work.
<gbruno> [github] t-paul pushed 124 additions 16 modifications 124 removals (Update more icons; Remove all manual dark mode handling. * Icons are changed based on the theme * Background colors are generated by blending a transparent overlay color (e.g. red) over the theme background color) https://github.com/openscad/openscad/commit/0a21c8939034c4510a146cdb59ac02a98912d793
<gbruno> [github] t-paul synchronize pull request #5563 (Add parameter list for local application start. Unify dark mode switch.) https://github.com/openscad/openscad/pull/5563
<JordanBrown> I have a dim memory of the test suite generating HTML output that presented expected and actual output side-by-side, et cetera. Was that just a dream?
<kintel> Oh, I guess ctest -C All includes test for existing bugs, which may include crash bugs
<kintel> Yeah, test suite outputs html, it should list an html file in the end
<kintel> But you need Python installed to generate it
little_blossom has quit [Quit: little_blossom]
<JordanBrown> Got Python.
<kintel> What does something as simple as "ctest -R echotest_echo-tests" output?
<JordanBrown> I guess if I want HTML I need to look to see what generates it, and work my way back from there to see why it isn't being generated.
<kintel> The ctest wrapper stuff is a bit non-obvious : (
<teepee> the html is mentioned at the end of the test run
<kintel> e.g. :
<kintel> Total Test time (real) = 0.39 sec
<kintel> Testing/Temporary/__apple-m1-pro_tfwc_report.html
<kintel> report saved:
<JordanBrown> https://bpa.st/WQZA
<JordanBrown> That's the output from that ctest invocation.
<JordanBrown> A long run with errors yields https://bpa.st/GWIQ at its tail.
<kintel> In your build folder, you should have sysinfo.txt and test_pretty_print.log.txt
<teepee> it did not find the python script then, I assume
<teepee> did you run ctest from inside build/tests ?
<JordanBrown> Yes.
<kintel> ^ we changed that recently. ctest should be run from the build folder
<JordanBrown> (Normally I run it using a wrapper script that does the cd for me.)
<kintel> *recently as in ~6 months ago or so
<teepee> yep, you can still run it there, but the report is only generated when run from "build" folder
<JordanBrown> ooh, when I run the little test run from the build directory, it does generate HTML.
<kintel> I bet it's your wrapper script
<JordanBrown> https://bpa.st/6NZQ
<kintel> Good, back in business :)
<JordanBrown> The wrapper script is pretty boring, but it did cd into "tests".
<JordanBrown> This is it: https://bpa.st/NUGQ
<JordanBrown> I need to do some tests to tune up the -j value for my "new" system.
<JordanBrown> Going back to that -C All run, linear_extrude-scale-zero-tests has a bunch of objects missing.
<JordanBrown> ERROR: The given mesh is not closed! Unable to convert to CGAL_Nef_Polyhedron.
<JordanBrown> Differences link: https://imgur.com/a/ceMVttH
<JordanBrown> Differences image, that is.
<JordanBrown> I can't immediately see how I could have affected linear extrude.
<kintel> Yeah, so just run ctest -C Good - those are likely known bugs
<JordanBrown> OK I'll try that.
<kintel> I haven't run that in a while, but running on now on master
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<JordanBrown> I just ran a default ctest run. Only one failure - 3mfexport_3mf-export failed because some XML file didn't have a proper CreationDate setting.
kintel has joined #openscad
<kintel> I saw teepee changed some CreationDate stuff recently for 3mf files
<JordanBrown> yes
<JordanBrown> Looks like %F and %T didn't work.
<JordanBrown> I assume that's an MSYS2 issue, though it seems strange.
<JordanBrown> example017 is failing. Looks kind of like a Z-fighting issue associated with the new color-preserving rendering.
<JordanBrown> https://imgur.com/a/y0g4Cmi is the expected-actual-mask images.
<JordanBrown> Loading it up into the program, the weird thing is that the tops of all three verticals show for F5, but only the two forward ones for F6.
<JordanBrown> To be precise, it was rendermanifoldtest-different_example017 that was failing.
<gbruno> [github] jordanbrown0 synchronize pull request #5553 (Start rotate_extrude on -X only for no angle specified.) https://github.com/openscad/openscad/pull/5553