-
Notifications
You must be signed in to change notification settings - Fork 151
Abort on startup (BAD LEXICON descriptor) #9
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
Thanks for trying and giving such a detailed bug report. Looking at this part
I conclude that something goes wrong with initialization of the 3d display pipeline of PythonOCC and then I do not handle it well later in my code. Could you give some details on your graphics card and the driver used? Also running
might be useful. |
GTX card is:
Some moaning about Vulkan in qtdiag - could be relevant?
|
Hi Adam. I have same issue, running in Manjaro Linux (Arch based) |
I don't really know why this works... it seems to be that the call to addLines() depends on things that havent been setup yet and delaying those callbacks to a later repaint fixes things. This should be investigated to determine the root cause.
I also ran into this issue on Arch Linux. I played around with it for a while and was able to narrow down a minimal change to get things to work. You can try it out at justbuchanan@cb316e0. I'm not familiar enough with the code to know why this fixes the problem and it's likely just a work around for some side-effects and not addressing the root issue. @adam-urbanczyk any thoughts on why this might fix the problem? Something else I've noticed that may or not be related is that a lot of "DEBUG:ipykernel.inprocess.ipkernel: ..." messages are printed out at startup that contain content from other ipython notebooks I've run recently, which are independent of cadquery/cq-editor. @jpmlt's log-runpy.log seems to contain similar content. |
Wow thanks. I'll check out your fix during the weekend. The logging noise is coming from the PythonOCC widget setting logging level to too low value AFAIK. I'll add an issue to track it. |
Hi Adam |
Great, please share your experience. I will check if the workaround is ok for CI too. |
@johnbeard does the workaround proposed by @justbuchanan work for you too? I investigated briefly and moving this line up self.components['object_tree'].addLines() actually prevents adding the helper X,Y,Z lines to the view. Somehow this addition seems to trigger the error (on your PCs, I cannot unfortunately reproduce it). @justbuchanan could you share what is your graphics card? I'm trying to find some common denominator here... |
I looked into this some more and found similar issues (tpaviot/pythonocc-core#279) reported in the pythonocc-core repo. It seems to be an issue with OCC finding data files that contain definitions of units for different locales. I got this working on the master branch (without the hack I posted above) just by setting the
Everything works great, including drawing of the helper axis lines! @adam-urbanczyk , did you already have |
@justbuchanan this is very interesting. I do not have If this is indeed the solution for everyone, I will change |
I'm looking at this again and the default value I have for Here are strace logs with Also output from |
The relevant lines from
|
It looks like the |
Great, getting to the bottom of it. Does it work if you unset the environment variable? |
Yes, if I unset the environment variable, everything works. Same for setting the variable to the correct path under The problem here seems to be that there's a bad interaction between Probably the best fix for myself and others with this problem would be to add |
OK, I just committed a workaround (run.py unsets @johnbeard @jpmlt could you confirm that this works for you too? @justbuchanan does this sound like Arch Linux bug to you? Any idea where to submit it? |
That workaround looks good to me 👍 I'm not sure if this is a bug in the opencascade Arch Linux package or not. I have version 7.3.0-3 (latest as of now) and it doesn't have the Lexi_Expr.dat file that cq-editor fails to find. This might just be an incompatibility between the latest versions of opencascade (which sets the env variable) and oce (which pythonocc uses). |
I looked quickly at the sources of 7.3 and the following fragment is quite revealing
So to my understanding @johnbeard @jpmlt would be great if you could confirm that this workaround is OK for you too. |
Below is what
|
Hi Adam Note that by unsettling CASROOT it runs even with the GTK issue, even if the result is ugly:
|
Great, good to hear! I'll think about permanently adding the |
I'm facing to a similar issue. When I execute the
(complete stacktrace here) I have the same error when I run the I also use Manjaro Arch, with the last version of CQ-editor, and the If I unset it, cq-editor doesn't start: the process is still alive, taking 100% of the cpu, but nothing append. I tried to export |
I vaguely remember seeing this before. Could you try compiling oce exactly like here: |
I get a crash when I start up on the current master (4e13229) on Arch:
GDB backtrace:
Conda setup:
Let me know if I can provide more details!
The text was updated successfully, but these errors were encountered: