stencil options

[wardini]

I am working with the stencil functions. It is very complicated to understand so I am not very confident I am working with it correctly. But it seems like the options "pass_op" and "depth_fail_op" are reversed in functionality. Is this possible? Thanks for any help.

[szabolcsdombi]

Here is an example which uses stencil:

zengl/examples/advanced/stencil.py

Lines 51 to 70 in 9002fbc

	stencil={
	'front': {
	'fail_op': 'replace',
	'pass_op': 'replace',
	'depth_fail_op': 'replace',
	'compare_op': 'always',
	'compare_mask': 1,
	'write_mask': 1,
	'reference': 1,
	},
	'back': {
	'fail_op': 'replace',
	'pass_op': 'replace',
	'depth_fail_op': 'replace',
	'compare_op': 'always',
	'compare_mask': 1,
	'write_mask': 1,
	'reference': 1,
	},
	},

If you worry about parsing and passing these values, here we do them (I checked for you they are in correct order in all places)

zengl/_zengl.py

Lines 551 to 567 in 9002fbc

	[
	True,
	STENCIL_OP[front.get('fail_op', 'keep')],
	STENCIL_OP[front.get('pass_op', 'keep')],
	STENCIL_OP[front.get('depth_fail_op', 'keep')],
	COMPARE_FUNC[front.get('compare_op', 'always')],
	int(front.get('compare_mask', 0xFF)),
	int(front.get('write_mask', 0xFF)),
	int(front.get('reference', 0)),
	STENCIL_OP[back.get('fail_op', 'keep')],
	STENCIL_OP[back.get('pass_op', 'keep')],
	STENCIL_OP[back.get('depth_fail_op', 'keep')],
	COMPARE_FUNC[back.get('compare_op', 'always')],
	int(back.get('compare_mask', 0xFF)),
	int(back.get('write_mask', 0xFF)),
	int(back.get('reference', 0)),
	]

zengl/zengl.c

Lines 1171 to 1186 in 9002fbc

	if (res->stencil_enabled) {
	res->stencil_front.fail_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_front.pass_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_front.depth_fail_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_front.compare_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_front.compare_mask = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_front.write_mask = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_front.reference = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_back.fail_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_back.pass_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_back.depth_fail_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_back.compare_op = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_back.compare_mask = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_back.write_mask = to_int(PyTuple_GetItem(settings, it++));
	res->stencil_back.reference = to_int(PyTuple_GetItem(settings, it++));
	}

zengl/zengl.c

Lines 867 to 875 in 9002fbc

	if (settings->stencil_enabled) {
	glEnable(GL_STENCIL_TEST);
	glStencilMaskSeparate(GL_FRONT, settings->stencil_front.write_mask);
	glStencilMaskSeparate(GL_BACK, settings->stencil_back.write_mask);
	glStencilFuncSeparate(GL_FRONT, settings->stencil_front.compare_op, settings->stencil_front.reference, settings->stencil_front.compare_mask);
	glStencilFuncSeparate(GL_BACK, settings->stencil_back.compare_op, settings->stencil_back.reference, settings->stencil_back.compare_mask);
	glStencilOpSeparate(GL_FRONT, settings->stencil_front.fail_op, settings->stencil_front.pass_op, settings->stencil_front.depth_fail_op);
	glStencilOpSeparate(GL_BACK, settings->stencil_back.fail_op, settings->stencil_back.pass_op, settings->stencil_back.depth_fail_op);
	self->current_stencil_mask = settings->stencil_front.write_mask;

• https://registry.khronos.org/OpenGL-Refpages/gl4/html/glStencilFuncSeparate.xhtml
• https://registry.khronos.org/OpenGL-Refpages/gl4/html/glStencilOpSeparate.xhtml

It seems we might be using glStencilOpSeparate incorrectly and indeed we swapped the [dpfail, dppass] parameters.
Is this what you are experiencing as the issue?

[szabolcsdombi]

would you like to create a PR with swapping the mentioned parameters at lines :873, :874 ?

[wardini]

Great. Thanks for finding this. I believe it is the source of the problem I was experiencing. I would do a pull request, but I actually don't know how to do it. Can you give me some time to figure it out and try?

[wardini]

OK. I tried to create one. Hopefully you can see it. I don't know how to perform tests on the source code so hopefully you can do that part. Thanks.

[szabolcsdombi]

tests will run on the MR, code looks correct. I will make a release. Thank you for finding and fixing this bug!

[szabolcsdombi]

2.7.2 is uploaded on PyPI and it includes your changes.
https://pypi.org/project/zengl/2.7.2/

pip install -U zengl