<kintel>
Yeah, turned out I ended up using both EGL and GLX :/
<kintel>
..but one day we can kill GLEW and things will be cleaner :) ..but need to wait until OpenCSG-1.6 is everywhere
califax has quit [Ping timeout: 252 seconds]
kintel has quit [Read error: Connection reset by peer]
califax_ has joined #openscad
aiyion has quit [Ping timeout: 252 seconds]
califax_ is now known as califax
aiyion has joined #openscad
kintel has joined #openscad
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
snaked has quit [Remote host closed the connection]
snaked has joined #openscad
ToAruShiroiNeko has quit [Ping timeout: 255 seconds]
ToAruShiroiNeko has joined #openscad
qeed_ has joined #openscad
qeed has quit [Ping timeout: 255 seconds]
<peepsalot>
what happened to ctest html reports?
<peepsalot>
no html generated anymore?
J23k60 has quit [Ping timeout: 245 seconds]
misterfish has joined #openscad
<peepsalot>
again, CI servers seem ok, i'm seeing build artifacts with html included, but local build is a different story. is there some extra parameter needed when running ctest ?
aiyion has quit [Ping timeout: 252 seconds]
<InPhase>
Hmm. All of the html outputs I have from the tests are from before I started running tests under the build directory.
<InPhase>
Coincidence or bug?
<InPhase>
I have 24 different copies of the repository stretched across the past 5 years, so that's at least a lot of different data points.
aiyion has joined #openscad
<peepsalot>
i have many under build/tests/ so i think that is coincidence
<InPhase>
Ok. The flip over poing for me was post 2021-02
<InPhase>
s/poing/point/
<peepsalot>
i'm showing some from 2023-01
aiyion has quit [Ping timeout: 252 seconds]
<peepsalot>
although I suppose file modified time doesn't necessarily correlate with source tree date
aiyion has joined #openscad
<InPhase>
I'll see if I can reproduce the old syntax I used to use to run them.
guso78k has joined #openscad
ToAruShiroiNeko has quit [Changing host]
ToAruShiroiNeko has joined #openscad
<guso78k>
teepee, i don't like my git history either, I have once taken the big procedure to re-copy openscad master to my work. it was big hazzle and had to disable all github security measures. you can see a branch 'master_old' as proof for that. Finally it did not solve, but I see squashed merges useful only. I don't see a reason to delete my fork and
<guso78k>
fork openscad from github again.
<guso78k>
breakpad is *the* state-of-the-art crash handler and "top dog". This depot_tool is really only needed to fetch the source code. further configure and make works totally without. Once a crash report is stored into a file, local tools can be used to create a readable stack trace.
<peepsalot>
why can't they use git to fetch source
<guso78k>
not sure , they are google.
<guso78k>
nobody dares to sync their code to github ;)
<peepsalot>
git != github
<guso78k>
sorry yes, yo are right ,
<guso78k>
however it appears that git is a subcommand of their depot_tool
guso78k has quit [Ping timeout: 245 seconds]
guso78k has joined #openscad
<InPhase>
peepsalot: Maybe it has to do with the fact that CTestCustom.template is set to only run the html output if cmake is less than version 2.8.12
mmu_man has joined #openscad
teepee has quit [Remote host closed the connection]
teepee has joined #openscad
mmu_man has quit [Ping timeout: 258 seconds]
mmu_man has joined #openscad
bozo16 has joined #openscad
cart_ has joined #openscad
guso78k has quit [Quit: Client closed]
bozo16 has quit [Ping timeout: 264 seconds]
arogora has joined #openscad
killjoy has quit [Ping timeout: 272 seconds]
arogora is now known as killjoy
killjoy has quit [Changing host]
killjoy has joined #openscad
mmu_man has quit [Ping timeout: 255 seconds]
<teepee>
ahem, google syncs the code to github themselves ;-)
snaked has quit [Remote host closed the connection]
snaked has joined #openscad
guso78k has joined #openscad
Freysgodi has joined #openscad
Freysgodi has quit [Client Quit]
mmu_man has joined #openscad
RoyK has joined #openscad
GNUmoon has quit [Ping timeout: 252 seconds]
snaked has quit [Ping timeout: 252 seconds]
GNUmoon has joined #openscad
guso78k has quit [Quit: Client closed]
guso78k has joined #openscad
<guso78k>
very good. did not know this link. at least this link was not included in the worksheet ;)
guso78k has quit [Client Quit]
guerd87 has quit [Read error: Connection reset by peer]
sinned69150 has quit [Quit: Ping timeout (120 seconds)]
rogeliodh has quit [Quit: Ping timeout (120 seconds)]
TheAssass1n has joined #openscad
TheAssassin has quit [Ping timeout: 252 seconds]
guso78k has joined #openscad
kintel has joined #openscad
<kintel>
peeps[work] fwiw, I haven't seen a change in HTML output; I've been using that on both Mac, Linux and Windows lately
<kintel>
..but we did make a change a few months back; you no longer have to run the tests from the "tests" folder. Running ctest from the build folder is the expected way
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<guso78k>
macdeployqt is a really great tool combining all necessary files into the AppImage and creating a Disk out of that. It copied all Python framework inside and it was almost working(i had to copy in the python environmental files manually there so it was working for william)
GNUmoon has quit [Ping timeout: 252 seconds]
GNUmoon has joined #openscad
rogeliodh has joined #openscad
<InPhase>
kintel, peepsalot: Well, I had commented out that if check for cmake version and set it running before going to sleep last night, but that didn't make the html output either. Not entirely sure what's missing here. But it is entirely possible that it stopped outputting for peepsalot and I on updating to 22.04? I so rarely look at the html output that I don't think I would have noticed this changing.
guso78k has quit [Quit: Client closed]
bozo16 has joined #openscad
kintel has joined #openscad
<kintel>
InPhase Huh, it's broken for both of you? I'm on 22.04 and it works perfectly here.
<InPhase>
Up-to-date 22.04.3 I assume?
<kintel>
uh, one sec
<kintel>
Ubuntu 22.04.2 LTS
<kintel>
What does your test log say?
<kintel>
running '/home/kintel/code/OpenSCAD/openscad/build/openscad --info' to generate sysinfo.txt
<kintel>
Post test:/usr/bin/python3 "/home/kintel/code/OpenSCAD/openscad/tests/test_pretty_print.py" --builddir=/home/kintel/code/OpenSCAD/openscad/build
<InPhase>
Where do you see that "--info" line? In LastTest.log?
<InPhase>
"--info" is not in mine.
<kintel>
On stdout when running tests
<InPhase>
Rerunning.
<InPhase>
Oh, for another comparison, are you using -j to parallelize testing?
<InPhase>
I haven't run without that in a while.
<InPhase>
I've been just... trusting it. :)
<kintel>
I usually do ctest -j 20, but in this case I did ctest -R cube and broke a cube test
<InPhase>
Okay, yeah, I'm consistently running with ctest -j20
<InPhase>
Is that --info line at the end? Because I don't have it.
<InPhase>
Nor at the top, so it's nowhere in output.
<kintel>
Both lines are just before starting to run the individual tests
<kintel>
Hmm, I did make some EGL-related changes to that file, could be my fault again :/
<kintel>
I'm running headless as usual
<InPhase>
This problem has existed for a while. Might still be your fault! ;) But if so from a while ago.
<InPhase>
I simply didn't notice it before, but I have some pretty old runs missing html output.
<kintel>
Could your run some limited test run, e.g. ctest -R cube and paste stdout somewhere?
sinned69150 has joined #openscad
<InPhase>
I completed a full run without -j as well, and still no post outputs.
<lf94>
What do you guys think of Blender as an "assembly" program, to assemble all the parts we make in OpenSCAD and other systems
<lf94>
Or do you think we should have something separate
<lf94>
that's simpler, and focuses on instancing and de-duplication
<lf94>
I feel OpenSCAD can serve as this
<lf94>
But it has too many extra features (like Blender)
<lf94>
And then the final output can be gltf or similar
<kintel>
I guess for assembly, we'd need metadata for individual parts to determine color/material, movement constraints etc.
<kintel>
..depending on what the purpose of assembly is of course, but I'd assume documentation generation, video/animation production etc.
<lf94>
metadata seems to be a large gap in our problem space
<lf94>
do we decouple metadata, or keep metadata local? hard to say
<lf94>
I think decoupling could make reasoning easier
<lf94>
You have your defined parts; then you specify where the parts should be, orientation, color, etc
_whitelogger has joined #openscad
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ccox has quit [Ping timeout: 255 seconds]
<InPhase>
kintel: Oh! Well that was actually it. Running in build "didn't work" because it was erroring out at the edit I just made, poorly, to CTestCustom.cmake. It did NOT error out when under build/tests because it was completely ignoring the CTestCustom.cmake file in build.
<InPhase>
kintel: So I reverted my edit to that file, and now it outputs html files if and only if I run ctest from build. If I run ctest from build/tests, the tests run but the html output is not generated. (pinging peepsalot)
<InPhase>
kintel: So we should phrase that not as "no longer a need" but that the old way is silently a bit broken. :)
<InPhase>
I never updated my testing method because it was locked down in a bash script and wasn't something I had to think about as easier or harder. When I went to do it manually I just did a cat of my bash script as a reference, and manually went to the same place I had in there.
<lf94>
I feel like you guys have been battling build issues forever
<InPhase>
And also it means we cannot run tests and get html without a build directory, because it failed for me similarly in <repository_top>/tests when I tried a run without using a build directory.
<InPhase>
lf94: That's a perpetual problem in modern cross-platform software.
<lf94>
Should just target Linux and Windows and call it a day
<InPhase>
lf94: For some languages it is more runtime issues. C++ migrates the majority of cross-platform issues to build time, which is an improvement, but still a source of frustration.
<lf94>
Better: just target Linux and get Windows people to maintain their own fork
<lf94>
You guys shouldnt be doing this for free
<InPhase>
Well it certainly is about time for someone to be sponsoring the project, but no entity with deep pockets has volunteered. :)
<lf94>
If Windows users still want to use OpenSCAD, tell them they gotta pay up with time or money
<lf94>
Same with mac
<lf94>
Same with *anything* really
<InPhase>
Even just 1-2 competent people full time on OpenSCAD development would really improve a lot of things I think, at the pace things are growing.
<InPhase>
There is stuff that just lags for a long time from the lack of this.
<lf94>
You guys could probably move faster only focusing on one OS
<lf94>
and then once you're happy with the large replacements, try for multi-OS again
<InPhase>
Well I doubt it.
mmu_man has quit [Ping timeout: 255 seconds]
<InPhase>
Our contributer pool actually sources from multiple preferred platforms, so keeping multiple OS's going adds contributing programmers, with a total contribution exceeding the labor cost of cross-platform build issues.
<InPhase>
Also, many contributers specifically look for "cross-platform" before deciding whether or not to contribute to a project, because it is a mark of the seriousness of a project and the likelihood of contributions to be broadly valuable and thus worth doing.
<InPhase>
Popularity of use matters for the longevity of a project, so indicators of that are part of an evaluation of a project.
<lf94>
interesting
<lf94>
for me I look: does it support Linux? no? im out
<InPhase>
I know I specifically checked for it in 2016 before deciding on OpenSCAD as a thing to focus on.
<lf94>
Going from Linux -> other platform is easier than the other way around from what Ive seen
<InPhase>
My hard line is does it support Linux. But additionally being cross-platform beyond that is pretty important even though I'm conscious of the frustrations that doing so introduces. :)
<InPhase>
And part of that is my prediction of my own usability of it in different scenarios. For example, I was running a course where I decided to introduce a 3D printing component. Well, I had 20 Windows laptops to distribute to student that I could do this with. So it was the Windows build of OpenSCAD (and Cura) that I had IT throw on those.
<InPhase>
s/student/students/
<InPhase>
Otherwise I would have had to choose between NOT doing a 3D component of the course, or figuring out how to do it with some other tool. And most likely another tool would not have achieved the same educational objective, because OpenSCAD puts the math of the geometry and of coordinate systems right in your face, and that was part of the learning objectives.
mmu_man has joined #openscad
J23k has joined #openscad
<lf94>
Hey, if you're the "Windows user", and you support the Windows, side, it's all good
<InPhase>
lf94: I'm personally more of a person who tries really hard to avoid touching Windows, but it's like sand, and it just gets everywhere sometimes.
<lf94>
I use it exclusively for gaming
<lf94>
Linux is good for everything else
<InPhase>
I found myself with minimal time for gaming since around age 18, so I just play some Linux games once in a while, and cross my fingers that Linux gaming will catch on fully before my retirement. :)
misterfish has quit [Ping timeout: 240 seconds]
guso78k has joined #openscad
guso78k99 has joined #openscad
<guso78k99>
l94 not sure if an engine which bases on SDF only can get as performant as CGAL only(unless SDF meshing gets a realy powerful algorithm)
guso78k99 has quit [Quit: Client closed]
TheAssass1n is now known as TheAssassin
_whitelogger has joined #openscad
mmu_man has quit [Ping timeout: 240 seconds]
<lf94>
guso78k: the idea would be to use both basically
<lf94>
one SDFs for complex shapes; use CGAL/Manifold for assembly
<guso78k>
lf94 i have seen your text , molten softly with a cube. so this used both ?
<guso78k>
the viewer should come up with a window, right ?
<lf94>
yea
<lf94>
nope, it used one :2
<guso78k>
your viewer did not recognize any 3d extension and quit
<lf94>
rip
<lf94>
lol
<guso78k>
is it theoretically possible to tell raylib to use a software renderer ?
<lf94>
maybe... you'll have to look
<lf94>
i would assume it could
<guso78k>
hmm, zig sounds very much like a big unkown to me, yet
guso78k has quit [Quit: Client closed]
kintel has joined #openscad
<kintel>
InPhase html report works for me, also without a build directory.
<InPhase>
Does it work in build/tests for you?
<InPhase>
If that's different, then we still have something to understand maybe... I would like to hear peepsalot's result.
<kintel>
Running from the tests folder yields the same issue as you; CTestCustom is not being picked up
<InPhase>
Ok. I do see you updated the docs already, so new devs hopefully won't do the wrong thing.
mmu_man has joined #openscad
<InPhase>
I'm not particularly worried about the outcome of running from <root>/tests either, because that's a terrible way to do things in comparison.
<InPhase>
I sort of only grandfathered that in for this repository for a phase in the past because of some prior build issues.
<InPhase>
I've also updated my testing script now.
<InPhase>
I guess if there's a way to make it more obvious that it's being done wrong if someone runs it the other older way, we could slide that in. Like if you ran it from tests it says, "Hey, wrong, do it one directory up".
<InPhase>
But on the whole I'm not too worried about that if we've documented the right path and we can inform the people who have been doing it the old now-wrong way and need to know.