"Shader Compilation Error" about skia-python

[pavpanchekha]

I imagine this is Skia's fault, not skia-python's; I'll try to narrow down the issue and if I can confirm that the issue is in Skia, I'll file a bug there.

[HinTak]

This has already been reported
I think: #214 - likely an upstream error - or lack of update of mac os x GL stack.

[HinTak]

It is probably not even skia's fault, but Apple's.

[HinTak]

Mac os x's GL stack is significantly behind others.

[HinTak]

I can't reproduce your shader problem on Linux. Not quite the same as #214 , as I see these errors instead:

$ python3 lab13.py http://browser.engineering/
OpenGL initialized: vendor=b'AMD',renderer=b'AMD Radeon R5 Graphics (radeonsi, stoney, LLVM 17.0.6, DRM 3.57, 6.8.4-200.fc39.x86_64)'
Script http://browser.engineering/feedback.js crashed SyntaxError: parse error (line 39)
    at [anon] (eval:39) internal
    at [anon] (duk_js_compiler.c:3797) internal
Script http://browser.engineering/book.js crashed SyntaxError: parse error (line 4)
    at [anon] (eval:4) internal
    at [anon] (duk_js_compiler.c:511) internal

BTW, your script relies on a specific way of installing dukpy - getting it from the Linux distribution (Redhat Fedora in my case), it is installed without any of the *.py scripts . Just dukpy.cpython-312-x86_64-linux-gnu.so . It looks like you need __init__.py, which loads from .evaljs import evaljs, JSInterpreter. I was seeing AttributeError: module 'dukpy' has no attribute 'JSInterpreter' error when I had it installed via the first method; and had to do pip install --user dukpy to get around it.

This may be a packaging bug on Redhat's part, in which case you (or I) should file at https://bugzilla.redhat.com/

[HinTak]

Let me know to retry if you can fix the crashed SyntaxError: parse error .

[pavpanchekha]

Ah, heh, the syntax error is unfortunately expected behavior! This is the browser running a website that uses JS syntax not supported by DukPy. Will look into RedHat

[HinTak]

I just get a window of your web site, and those messages. So i assume that's correct behavior? In that case this is mac os x specific and same as #214 .

As for the packaging issue, the file listing of the package is https://koji.fedoraproject.org/koji/rpminfo?rpmID=35288394 - missing all the *.py files. I am not familiar with it, and how it should be.

[HinTak]

Btw, the lab13 script also seems to be doing a busy wait and tie up one of my cpu's while it runs. (Just looking at my system load monitor).

[pavpanchekha]

I’ll look, may be an issue

[HinTak]

Argh, the busy wait loop is the while True: on line 1491 in mainloop. Polling for events isn't ideal - there has got to be a way of registering event handler, for better efficiency.

[pavpanchekha]

Closed as duplicate