03:46 <pca006132> J24k27: no, I just mean undefined variables will not cause error

03:46 <pca006132> iirc you talked about this issue a few weeks ago, and I cannot reproduce it?

05:45 <J24k27> pca006132 i am using 2024.07.27 (after you told me you don't have this issue)  - now i wonder do you use windows, or do you have a much faster CPU? how long does ```for (i=[0:100],j=[0:100]) rotate(i*j+a) cube();``` take

05:46 <pca006132> I am using Linux, and the expression took quite a while

05:46 <pca006132> but the window is still responsive, and messages were printing in the console

05:46 <pca006132> I don't think there is a way I can stop it though

05:49 <J24k27> at least on windows it is so frozen that even if i copied the code into the clipboard - i can not paste it into a new second openSCAD process

05:50 <J24k27> and it is frozen for several minutes - this seems to be exponential with the loop and when working with pattern, every typo needs to kill the software

05:52 <J24k27> using "stop at first warning" works - but as manifold produces warnings this can't be used at the moment - and if using points this is also not working

05:55 <J24k27> pca006132 i would totally believe that windows is the culprit here - however a lot users use openSCAD on windows

05:55 <pca006132> I see

05:55 <pca006132> maybe some Qt thing I guess

05:56 <J24k27> didn't there was an update of the Qt version?

06:07 <pca006132> Not sure if Qt6 is used by default now in snapshot builds

06:07 <pca006132> at least I am still building with Qt5 locally

06:08 <guso78> You can Open a New openscad and Paste in the content of the Backup File

06:10 <guso78> I think openscad should Always Stop on undefined vars. Intended undefined should be assigned explicitely 😊

06:11 <InPhase> guso78: At the very least the first time undefined propagates into a geometry object this would be a reasonable stopping point.

06:11 <InPhase> It will never work.

06:13 <InPhase> I cannot think of a single design in production where part of the model failed to render but people still considered this okay and usable. Nobody is out there making models where partial failure to render is okay.

06:13 <InPhase> So we don't lose anything by giving up on pretending this is okay. :)

06:17 <pca006132> I thought openscad should automatically backup before render? in case it will crash during render

06:20 <InPhase> It does if and only if you have not saved first.

06:20 <J24k27> guso78 yes you need to use the backup (which could be presented in the menu or splash screen) - i was referring to the (windows?) issue that the clipboard doesn't work if the process where the copy originated is unresponsive - but i can copy the code into a notepad and then copy again and paste into openSCAD. Else i have two frozen processes

06:24 <J24k27> inPhase i guess people ignore some warnings if they don't have influence on the outcome like cylinder(10,d=10,d1=15);

06:25 <J24k27> or if your loop producing cube(0); at first

11:18 <guso78> J24 offering Backup Versions would indeed be an improvement, but there might be several Versions to choose from... And it would Show weakness to the User. Not an Option for evaluated Python as this Runs in a separate Thread anyway

11:20 <guso78> Pca yes, Always Backup even before parsing the Code!

11:20 <J24k27> just put the backup folder beside the "example" - or even offer to open the backup " there is a unsaved design you like to restore it"

11:21 <J24k27> and i like a longer "recent" list (or customize the number of entries)

11:27 <guso78> Yes i will consider the Option to Implement loading of Backup Files

11:27 <guso78> . Should Not be a Security Hazard when loading and parsing *any* Files will Not create Buffer overflows...

11:30 <J24k27> haha just  noticed other gits like cura or bambu - just putting issues to inactive if nobody responds to the bug - Ü, There is no problem if we all ignore it long enough.

21:45 <gbruno> [github] praetp opened issue #5253 (Openscad crash ) https://github.com/openscad/openscad/issues/5253