<peepsalot>
haven't run a local build + tests in a while but seems I'm getting 1237 tests failed
<peepsalot>
GLEW Error: Unknown error
<peepsalot>
Can't create OffscreenView: Unable to initialize Glew.
<peepsalot>
guess i'll check if reboot helps
peepsalot has quit [Quit: Connection reset by peep]
peepsalot has joined #openscad
<peepsalot>
no luck after reboot, thought maybe nvidia drivers updated and broke things again
qeed has joined #openscad
qeed_ has quit [Ping timeout: 260 seconds]
hyvoid has quit [Ping timeout: 255 seconds]
hyvoid has joined #openscad
hyvoid has quit [Ping timeout: 272 seconds]
hyvoid has joined #openscad
guso78k has joined #openscad
guso78k has quit [Quit: Client closed]
<peepsalot>
i guess linux CIs are doing ok. are tests working for any other linux users on normal desktops (non-headless)? CI uses some special xvfb setup iirc?
guso78k has joined #openscad
aiyion has quit [Remote host closed the connection]
aiyion has joined #openscad
peeps has joined #openscad
peeps[work] has quit [Ping timeout: 255 seconds]
peeps has quit [Ping timeout: 240 seconds]
peeps[work] has joined #openscad
RoyK has quit [Ping timeout: 240 seconds]
<Scopeuk>
not sure if anything has changed with kintel's recent work but the ci used to create a virtual monitor in software to draw onto
<peepsalot>
that's what i mean about xvfb (X virtual frame buffer). i'm wondering if any systems *not* on xvfb are working after the latest gl context changes
<peepsalot>
the gui works fine, what fails is the offscreen context, which is only triggered from command line png export afaik
teepee_ has joined #openscad
teepee has quit [Ping timeout: 252 seconds]
teepee_ is now known as teepee
teepee has quit [Remote host closed the connection]
<guso78k>
without command line png export most of the ctests would fail ...
<peepsalot>
guso78k: yeah, 23% tests passed, 1237 tests failed out of 1616
<peepsalot>
i know, i acknowledged that linux CI servers are reporting success
<peepsalot>
anyways, its past 3am here, i should get to bed.
<teepee>
good night, we can have a look later, default builds may now use EGL but that should be supported by nvidia too
ToAruShiroiNeko has quit [Ping timeout: 255 seconds]
ToAruShiroiNeko has joined #openscad
<guso78k>
teepee, python-pr3 got so many changed files again. is it useful trying to split off preparation stuff only ?
<teepee>
number of files is not a problem
<teepee>
those CI errors were mostly some checkout problem
<teepee>
master is fine except the collition of manifold and gcc-13 in mxe-64bit which I hopefully find the time to fix sometime
<teepee>
actually maybe lets still push the gcc-11 for the meantime
cart_ has joined #openscad
ToAruShiroiNeko has quit [Ping timeout: 248 seconds]
ToAruShiroiNeko has joined #openscad
arogora has joined #openscad
killjoy has quit [Ping timeout: 252 seconds]
arogora is now known as killjoy
killjoy has joined #openscad
killjoy has quit [Changing host]
ToAruShiroiNeko has quit [Ping timeout: 255 seconds]
ToAruShiroiNeko has joined #openscad
guso78k has quit [Quit: Client closed]
mmu_man has joined #openscad
guso78k has joined #openscad
misterfish has joined #openscad
<InPhase>
peepsalot: xubuntu 22.04.03 LTS, Ran tests on my 2023-09-17 build, 100% pass. Pulling and clean building...
<InPhase>
peepsalot: Hmm, no, that's a disaster. Fresh build after a 2023-10-10 pull says 23% pass now.
<InPhase>
So that was a code change between 2023-09-17 and 2023-10-10 that breaks nearly every test.
<InPhase>
How on Earth did we manage to do that?
<InPhase>
"GLEW Error: Unknown error" and "Can't create OffscreenView: Unable to initialize Glew." thoughout the testing log.
<InPhase>
Looks like all changes between these dates were kintel.
* InPhase
draws a circle, waves his magic wand, and summons kintel.
hyvoid has quit [Quit: WeeChat 4.0.5]
hyvoid has joined #openscad
<InPhase>
Ah yes, including two commits specifically about GLX and offscreen support.
<InPhase>
Or, three.
<InPhase>
Okay, and it's not the testing framework changes. A manual invocation of "openscad -o foo.png foo.scad" on the master branch fails with the same error. So we broke all non-gui usage.
kintel has joined #openscad
<kintel>
teepee could you remind me what's needed to get rid of the 64-bit MXE build failure?
<kintel>
uh, nvm, just read the history :)
<kintel>
peepsalot: Please hit me up when you have a moment; this very much sounds like my domain, and I'd like to understand what happens before you magically make it work again :)
<kintel>
Just to verify, could you try "cmake -DUSE_BUILTIN_OPENCSG=ON" and verify that it works?
<InPhase>
Rerunning with that.
kintel has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kintel has joined #openscad
kintel has quit [Client Quit]
J23k32 has quit [Quit: Client closed]
J23k32 has joined #openscad
<teepee>
oh, haha, opps: 6,2 GB of 2 GB used
<teepee>
argh
<teepee>
rebuilding the MXE images is good, checking in the changes to actually go back to gcc-11 would be much better though
kintel has joined #openscad
<kintel>
InPhase If you feel like testing more: cmake -DENABLE_EGL=OFF
<kintel>
Btw., you do have a real display right? ..with DISPLAY set etc.?
teepee_ has joined #openscad
<InPhase>
kintel: Correct, DISPLAY is :0.0 and works.
teepee has quit [Ping timeout: 252 seconds]
teepee_ is now known as teepee
<InPhase>
And I can run another test in the background. My clean build and test process is all scripted, all the way down to a special kinteltest script, so it's no burden to keep running conditions. ;)
<kintel>
cool, let me know how ENABLE_EGL=OFF goes
<InPhase>
Passes.
<InPhase>
That's -DENABLE_EGL=OFF but after removing the -DUSE_BUILTIN_OPENCSG=ON
<InPhase>
Technically I used cmake -DEXPERIMENTAL=ON -DCMAKE_BUILD_TYPE=Release -DENABLE_EGL=OFF
<InPhase>
The first two I always have set.
<kintel>
I wonder if it's a xubuntu-specific issue. When I have some time I'll test in a xubuntu Docker container
<kintel>
My guess is that this has something to do with the EGL->GLX fallback code
<InPhase>
I thought peepsalot was using Mate or something.
<InPhase>
From one of those other distros...
<kintel>
humm, and two different GPUs..
<kintel>
I haven't tested this myself on a real display, just using ssh. I'll try to test that myself as well.
<InPhase>
Is this EGL thing expecting some sort of Wayland background?
<InPhase>
Because I do not have that...
<kintel>
no, EGL is independent of windowing systems
<kintel>
Also, when building with -DUSE_BUILTIN_OPENCSG=ON, we'll use EGL everywhere
<InPhase>
Oh. Well that sounds like a weird combination of results then.
<kintel>
..because old OpenCSG uses GLEW and hence requires GLX on most systems
jonasbits has quit [Quit: No Ping reply in 180 seconds.]
jonasbits has joined #openscad
<kintel>
Nice, looks like the Dutch people are in charge now :)
<kintel>
Btw., didn't youmagine.com follow Ultimaker into the new company?
<teepee>
no idea, I think it's separate now, but I did not really have a closer look
<teepee>
still a bit strange all that shuffling :)
<teepee>
so now it's the big 3? Thingiverse, Printables and the new kid on the block MakerWorld
<teepee>
well, plus the other ones with more focus on paid stuff, but I do have the feeling that's still not working too well
<teepee>
I guess it would be nicer if that worked, having the places at least separate from the printer builders
<InPhase>
kintel: Excellent, glad you were able to reproduce it locally.
<teepee>
also mxe64 is back on gcc-11
<peeps[work]>
today i learned twitter hides all conversation if you aren't logged in. can't see that the OpenSCAD tweet is a reply to anyone, and can't see any replies to OpenSCAD's tweet
<teepee>
yep, right after they blocked all access for a while breaking every newspage doing includes
<InPhase>
Yeah, if there were replies to that tweet which provided information, I couldn't see them.
<InPhase>
I delayed getting a twitter account because I didn't like the general sentiment of twitter behavior, and now I certainly have no reason to create an account with them appearing to fall apart at the seams. Them breaking the site more is not going to motivate me to register. :)
<teepee>
mr. technoking is contemplating forcing paid accounts, that will make it an instant success and source of piles of money, I'm sure
<teepee>
of course he's doing that only for the good of mankind as he has no interest whatsoever in money for twitter
J23k32 has quit [Quit: Client closed]
J23k32 has joined #openscad
aiyion has quit [Ping timeout: 252 seconds]
aiyion has joined #openscad
prev has joined #openscad
<teepee>
guso78: nettle almost done for MXE
prev has quit [Quit: Client closed]
prev has joined #openscad
<prev>
Anyone know what change was made that bumped the minimum OS X requirement to 10.14? I'm trying a 10.13 build from source but I'm not sure what the breaking change actually was.