GNUmoon has quit [Remote host closed the connection]
GNUmoon has joined #openscad
snaked has joined #openscad
snaked has quit [Quit: Leaving]
snaked has joined #openscad
mmu_man has quit [Ping timeout: 252 seconds]
AzAv8r has joined #openscad
AzAv8r has quit [Client Quit]
AzAv8r has joined #openscad
teepee has quit [Remote host closed the connection]
<AzAv8r>
Help! Using the 4-29 nightly snap on Ubuntu 24.04, the editor has suddenly occupied full window-width with no means to restore it to the left pane. I can only enable/disable it, and it always will occupy full-width of the Openscad window. I have had other problems with the editor (it gets confused about syntax highlighting after I keep it open
<AzAv8r>
for a few days, but that always goes away if I exit and restart. The new problem persists through restarts. Any ideas? :(
teepee has joined #openscad
<InPhase>
AzAv8r: First try the latest AppImage from the download page.
<InPhase>
AzAv8r: If this exhibits the same problem, then there's a problem with code + your system.
<InPhase>
If it only hits with the snap, then something is broken on the snap. I don't know who is packaging up the snap you grabbed.
<InPhase>
teepee: Oh, well we link to the snap. Are we actually auto-releasing the snap?
<AzAv8r>
Snap is in the "nightly - stable" Ubuntu Store channel (there is also a "nightly-dev", Iirc.
<AzAv8r>
This came via the Ubuntu Store (or whatever they are calling it these days...)
<InPhase>
AzAv8r: I just tested the latest AppImage, 2025-05-07, and see no issues.
<AzAv8r>
Feels like a QT settings issue - it's persistent.
<InPhase>
There is also the solution path of just deleting your openscad config file.
<InPhase>
It will wipe out your settings, but some of those settings have gui elements.
<AzAv8r>
Happened after several hours of use. Certainly could be a GNOME issue - that seems to have been relatively unstable in the last several Ubuntu updates (even if they are LTS)
<InPhase>
You could also try to selectively delete the gui setting bit.
<InPhase>
AzAv8r: In ~/.config/OpenSCAD/OpenSCAD.conf at the bottom under [window] there is some @ByteArray crap. Sometimes this goes bad.
<InPhase>
It's qt layout data.
<InPhase>
In particular, state= is stuff like where to put the resizeable subcomponents like the editor.
<InPhase>
You'd need to close all running instances, then delete it, then open again.
<InPhase>
Probably best to delete both of those lines just to be safe.
<InPhase>
And probably it would be more Human Friendly if we had a Window, Reset Defaults from the menu, that basically did the equivalent of this.
<AzAv8r>
Ah, just looked at that file and see the two arrays under "Window". Will delete those.
<AzAv8r>
"geometry" and "state" - both @ByteArray
* InPhase
plays suspenseful music.
<AzAv8r>
Yikes. Deleting both geometry and state resulted in no change when I restarted OpenSCAD. :{
<AzAv8r>
I'm going to reboot again - just because. Be back momentarily.
AzAv8r has quit [Quit: Client closed]
AzAv8r has joined #openscad
<AzAv8r>
Back. No joy.
<InPhase>
Did you close the program before deleting those entries?
<InPhase>
It will save its existing state every time you close it. So delete, close, reopen, is like never having deleted.
<AzAv8r>
Yes. And entries under [window] are blank.
<AzAv8r>
I'm not particularly familiar with Qt, but I'd expect sizers I could grab at the pane (window) boundaries to resize. The editor only does that at the bottom (boundary between editor pane and 3d view pane), nothing on the sides. Interesting, just noticed it also does not show up on the "task bar" (not sure what Ubuntu calls it; left side where
<AzAv8r>
pinned and open programs show up.) I has always showed up there before.
<AzAv8r>
Iirc Ubuntu pushed some update today - maybe we (I) should assume this is a bad Ubuntu patch, even thought it is on the current LTS branch?
<InPhase>
Yes, there are sizers you can grab when things work right.
<InPhase>
The editor can be freely resized left and right.
<InPhase>
I'm on xubuntu, to avoid the instabilities that gnome had way back when I made this choice.
<InPhase>
That and I never liked that side bar approach.
<AzAv8r>
Okay, not working. Let's run with the assumption for now that this Ubuntu update broke something. I've had a suspicion for the last week or so that there was some Gnome stuff going on with Ubuntu which was breaking things. The main thing I noticed was that "print screen" (for a screen snap) worked differently. And today it started
<AzAv8r>
working the way I was used to it working. Hmmm. Somebody is playing Wack-a-mole...
<InPhase>
And did you test the AppImage?
<InPhase>
There's no install for those. You just run it.
<InPhase>
chmod +x it, and then execute it.
<AzAv8r>
I can - give me a few. Does it use the same config files, or should I somehow sandbox it?
<InPhase>
It uses the same user level configs.
<InPhase>
No need to sandbox it.
<InPhase>
It autobuilds as a git process, so the security level is on par with the repo itself.
<AzAv8r>
Okay - appimage works properly - display I'd expect. To be clear, that is dated 5/7 vs 4/29.
<InPhase>
Hmm.
<InPhase>
Well, good! But, why? :)
<AzAv8r>
Yep = somehow isolated from the native environment. Don't know enough about Appimage vs SNAP to understand what this could mean.
<InPhase>
There's a "Help, Library Info" box.
<InPhase>
If you can access that in both of them, and save that to two text files, you can diff them.
<InPhase>
I am assuming from your prior comments you know diffing. If not, let me know. :)
<AzAv8r>
Appimage: OpenSCAD version 2025.05.07.ai25355 (git 72c9919d6)
<AzAv8r>
Cool. It's a good day (every day I learn something. Which I hope is every day...)
<InPhase>
Interesting, clang vs gcc, but I don't think that should break it.
<InPhase>
User Config Path: /home/jon/snap/openscad-nightly/5069/.config/OpenSCAD
<InPhase>
It's not even reading the ~/.config/OpenSCAD for snap.
<InPhase>
So you would have to edit this separate location to try to fix snap...
<AzAv8r>
Wilco
<InPhase>
Yep. Those are my four guesses. Most likely to least likely: 1. The other config file. 2. GLib 2.80.0. 3. Qt 5.15.13 (should be fine though). 4. Clang 18.1.3.
<InPhase>
Or, some wacky security setting for the snap.
<InPhase>
But then I would not have expected it to appear as a problem in the middle of working.
<AzAv8r>
Bingo! Deleting the two byte arrays under "Window" in the Snap Config file fixed the problem. I did notice one of the arrays (second one - state?) had quotes around it (the values - " "), vs the first one (window?) did not. Unfortunately, I simply deleted them rather than saving a copy first. Once upon a time I had a photographic memory -
<AzAv8r>
that is not now :(
<AzAv8r>
And the nightly now shows up in the "task bar", or whatever Ubuntu calls it.
<AzAv8r>
In the new "as working" world, the "window" block in the snap/OpenSCAD/Openscad_config file ONLY contains the geometry bytearray, and it is without quotes.
<AzAv8r>
I should have saved that file - it would have been useful for further analysis. This is why I needed to retire...
<InPhase>
Interesting. The quotes are a weird defect.
<InPhase>
Maybe snap updated something and put them there?
<InPhase>
I don't see how openscad itself could have.
<InPhase>
Well, I'm glad it turned out to be a rational and comprehendible issue in the end, even if the trigger for it remains a mystery.
<InPhase>
If you end up spotting some pattern that triggers it, that could be helpful to raise.
<AzAv8r>
Thanks for your assistance. I shall attempt to be more observant as anomalies arise such that I can provide better data.
<AzAv8r>
The one pattern I have consistently been observing lately is the editor getting "confused" with respect to syntax validation - I'd see an indicator that I had not closed a scope (the line on the left side of the editor continuing to the bottom of the file...) and yet a test of every scope closure element (curly bracket, semicolon, or parenthesis)
<AzAv8r>
would show proper closure per the syntax highlighter. Sometimes if I scrolled above the start of the problem and made an edit (added/deleted a meaningless character like a space), then scrolled back down, the closure error would magically disappear. I cannot say this always worked.
<AzAv8r>
In all cases before, when I had odd behavior, I'd save the file, exit Openscad and restart it. I do have the release copy also, but as I *really* like current features like Manifold and the PDF dialog features I generally use the nightly. And I can say I've only noticed anomalies in the last month or two. Presumably (I haven't looked)
<AzAv8r>
the editor is another project, so an error there could explain the syntax validation issue I have observed.
<AzAv8r>
Again thank you. I need to drop off for the night (GMT-7).
AzAv8r has quit [Quit: Client closed]
rawgreaze has quit [Ping timeout: 265 seconds]
rawgreaze_ has joined #openscad
rawgreaze_ is now known as rawgreaze
howiemnt has quit [Remote host closed the connection]
howiemnt has joined #openscad
stealth_ has quit [Quit: Leaving]
mmu_man has joined #openscad
rogeliodh has quit [Ping timeout: 252 seconds]
rogeliodh has joined #openscad
snaked has quit [Quit: Leaving]
Guest74 has joined #openscad
Guest74 has quit [Client Quit]
LordOfBikes has quit [Ping timeout: 265 seconds]
LordOfBikes has joined #openscad
kintel has joined #openscad
<kintel>
In terms of window geometry woes, we should probably either add some reset feature, or auto-reset old configs at some point; I believe an incompatibility was introduced at some point to fix a bug, but I cannot remember if it was Qt5 vs. Qt6 or something else..
* InPhase
nods.
<InPhase>
The reset sounds important. It doesn't rise as an issue too often, but when it does, it's out of scope for mortals to diagnose and self-fix.
<kintel>
But in any case, this issue has come up a handful of times the last few months
<InPhase>
Hmm. Perhaps some version of something is triggering a problem at a faster pace.
<kintel>
..and on some platforms (like macOS), the config file is also cached, so editing it isn't always enough : /
<InPhase>
Oof. Weird.
<InPhase>
Is there a way to tell it to always read fresh from disk?
<kintel>
I assume last mile app startup optimization or smth.
<kintel>
You kill the cache server
<InPhase>
I mean within code.
<kintel>
phew - no idea, hidden inside layers of Qt
<InPhase>
Like, disable the cache.
<InPhase>
I'm pretty sure caching of the config file is not meaningfully improving our startup times. :)
<nomike>
First of all: It's missing the "REQUIRED" flag, so no wonder the build didn't fail. And secondly: Why is there a hard dependency on version 3.4?