whitequark changed the topic of #solvespace to: SolveSpace--parametric 2d/3d CAD · latest version 2.2 · http://solvespace.com · code at https://github.com/solvespace/solvespace · logs at https://irclog.whitequark.org/solvespace
u2fly has joined #solvespace
<u2fly> Hello
u2fly has left #solvespace [#solvespace]
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit I figured out why I had to do glFrontFace. This is because when we do `UpdateProjection(/*flip=*/true)`, our projection matrix inverts the handedness of the coordinate system. And this worked in OpenGL 1 because it does culling with modelview matrix, and in our case it does culling with modelview*projection matrix. https://github.com/solvespace/solvespace/issues/102#issuecomment-2614
<cr1901_modern> Why would OpenGL (even 1) do culling before going to projection space?
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit I've updated the opengl2 branch. There's one last bug, back faces aren't actually drawn for some reason... https://github.com/solvespace/solvespace/issues/102#issuecomment-261442200
<whitequark> cr1901_modern: dunno?
<cr1901_modern> Actually, I just remembered something. In OpenGL 1 original version of solvespace, jweustes doesn't actually ever do culling: http://solvespace.com/forum.pl?action=viewthread&parent=961 (last reply). Did that code change to necessitate culling?
<whitequark> eh sorta
<cr1901_modern> In any case, that comment just caught my attention. The three.js viewer doesn't cull either. I'll take a look tomorrow at that issue b/c I'm curious how your new code works.
<travis-ci> solvespace/solvespace#205 (opengl2 - d23cc88 : whitequark): The build passed.
<GitHub> [solvespace] Evil-Spirit commented on issue #102: @whitequark,... https://github.com/solvespace/solvespace/issues/102#issuecomment-261446068
<travis-ci> solvespace/solvespace#206 (opengl2 - f8065d3 : whitequark): The build was broken.
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined #solvespace
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit I've already fixed that actually https://github.com/solvespace/solvespace/issues/102#issuecomment-261452521
<travis-ci> solvespace/solvespace#206 (opengl2 - f8065d3 : whitequark): The build passed.
<GitHub> [solvespace] Evil-Spirit commented on issue #102: @whitequark, I mean "back faces aren't drawn, even when enabled" https://github.com/solvespace/solvespace/issues/102#issuecomment-261455858
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit Ah yeah, please look into it https://github.com/solvespace/solvespace/issues/102#issuecomment-261458802
mumptai has joined #solvespace
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit Another problem: on OS X with 3.2 core profile enabled, GL returns 0x502 at the following line in MeshRenderer::Draw:... https://github.com/solvespace/solvespace/issues/102#issuecomment-261476404
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit OK I figured it out, GL 3.2+ Core Profile requires VAOs to be used (https://stackoverflow.com/questions/24643027/opengl-invalid-operation-following-glenablevertexattribarray, https://stackoverflow.com/questions/5970087/understanding-vertex-array-objects-glgenvertexarrays). On the other hand, OS X in Compatibility Profile allows *only* `#version 120` shaders to be used.... https://gi
<GitHub> [solvespace] Evil-Spirit commented on issue #102: @whitequark,... https://github.com/solvespace/solvespace/issues/102#issuecomment-261478780
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit We are, but OS X cannot render GLES 2.0, so workarounds are needed. But I have thought up how to do it in only a mildly horrible way. https://github.com/solvespace/solvespace/issues/102#issuecomment-261479046
<GitHub> [solvespace] Evil-Spirit commented on issue #102: What about Angle for OsX? https://github.com/solvespace/solvespace/issues/102#issuecomment-261479241
<GitHub> [solvespace] whitequark commented on issue #102: It's possible but I would like to avoid doing all this more complicated than it already is... https://github.com/solvespace/solvespace/issues/102#issuecomment-261480043
<whitequark> dear lord
<whitequark> I hope I have sinned very badly in my previous life because otherwise I did not deserve OpenGL
<ohsix> why?
<whitequark> there are three main components in OpenGL
<whitequark> suffering, suffering and more suffering
<ohsix> ya i know, just wondering what this one is
<whitequark> let me commit something
<ohsix> there are some areas where the genobject idioms don't match up, like vbos/vertex arrays, and the disconnect is really jarring
ohsix_ has joined #solvespace
ohsix_ is now known as nohsix
<nohsix> ;]
<nohsix> eta? only got a bit
<whitequark> in a moment
<nohsix> 👍
nohsix has quit [Quit: Lost terminal]
<whitequark> ok it seems to work
nohsic has joined #solvespace
<nohsic> whoops
nohsic is now known as nohsix
<nohsix> restarted ibus and gnome-terminal crashed
* nohsix needs to remember that specific peril
<GitHub> [solvespace] Evil-Spirit commented on issue #102: @whitequark,... https://github.com/solvespace/solvespace/issues/102#issuecomment-261488470
<GitHub> [solvespace] Evil-Spirit commented on issue #102: @whitequark,... https://github.com/solvespace/solvespace/issues/102#issuecomment-261488470
<nohsix> ahhh
<nohsix> nice
<nohsix> you might want to look at ANGLE
<nohsix> i mean, it sucks to have to use it for something like this, but it is to solve those kinds of problems
<travis-ci> solvespace/solvespace#207 (opengl2 - a11f267 : whitequark): The build passed.
<nohsix> you target gles and it adapts to different gl versions and to directx where needed
<nohsix> oh, you did mention angle
<nohsix> 'good' support of gl always seemed to include some craziness, though it's more interesting with the core profiles and es
<nohsix> doom3 had like 6 main rendering pipelines for different feature levels of shaders, down to fixed function
<nohsix> k gotta sleep, adios
nohsix has quit [Quit: leaving]
mumptai has quit [Remote host closed the connection]
<GitHub> [solvespace] whitequark commented on issue #102: @Evil-Spirit I fixed the back face problem. It was because SSurface::TriangulateInto first filled the mesh with triangles with empty color, and then filled the color in separately. https://github.com/solvespace/solvespace/issues/102#issuecomment-261503165
<GitHub> [solvespace] whitequark pushed 1 new commit to 2.x: https://github.com/solvespace/solvespace/commit/ece763062489996a4042220eaaf90338353c4be5
<GitHub> solvespace/2.x ece7630 whitequark: Fix detection of transparent meshes....
<travis-ci> solvespace/solvespace#209 (opengl2 - ebf84f4 : whitequark): The build passed.
<GitHub> [solvespace] whitequark pushed 3 new commits to master: https://github.com/solvespace/solvespace/compare/e56630e71d2a...9a0f2c36013f
<GitHub> solvespace/master 73844f7 whitequark: Draw triangle back faces like the front, unless we draw them in red....
<GitHub> solvespace/master 9a0f2c3 whitequark: Fix detection of transparent meshes....
<GitHub> solvespace/master c8ff17f whitequark: Add OpenGL 2 support on Windows using ANGLE....
<GitHub> [solvespace] whitequark commented on issue #102: this is finally **MERGED** https://github.com/solvespace/solvespace/issues/102#issuecomment-261512667
<GitHub> [solvespace] whitequark commented on issue #89: Fixed in master https://github.com/solvespace/solvespace/issues/89#issuecomment-261512708
<travis-ci> solvespace/solvespace#210 (master - 9a0f2c3 : whitequark): The build passed.
<GitHub> [solvespace] whitequark opened issue #114: Flickering when resizing a window https://github.com/solvespace/solvespace/issues/114
<GitHub> [solvespace] whitequark opened issue #115: Normals are see-through https://github.com/solvespace/solvespace/issues/115
ohsix has quit [Ping timeout: 250 seconds]
ohsix has joined #solvespace
<GitHub> [solvespace] whitequark commented on issue #113: @Evil-Spirit yes, please look into it. This seems easy enough and potentially very useful. https://github.com/solvespace/solvespace/issues/113#issuecomment-261526460
<GitHub> [solvespace] whitequark commented on issue #21: @Evil-Spirit please implement this check https://github.com/solvespace/solvespace/issues/21#issuecomment-261528279
<GitHub> [solvespace] Evil-Spirit commented on issue #115: @whitequark, yes, we have it everywhere. probably this is because tricks with depth buffer in ssglDepthRange https://github.com/solvespace/solvespace/issues/115#issuecomment-261534105
<GitHub> [solvespace] whitequark commented on issue #115: @Evil-Spirit I'm not sure. This is not a problem in version 2.0, 2.1 or 2.2, but it is a problem with 3.0 with OpenGL 1 and OpenGL 2. It was introduced somehow during the Canvas changes I think. https://github.com/solvespace/solvespace/issues/115#issuecomment-261535408
<GitHub> [solvespace] whitequark commented on issue #115: But 2.0 still uses the ssglDepthRange hacks. https://github.com/solvespace/solvespace/issues/115#issuecomment-261535442
<GitHub> [solvespace] whitequark pushed 1 new commit to master: https://github.com/solvespace/solvespace/commit/0ae0343fc79ea9510a0fb080c0ff1e050185d4f0
<GitHub> solvespace/master 0ae0343 whitequark: Add a simple benchmarking harness.
<travis-ci> solvespace/solvespace#211 (master - 0ae0343 : whitequark): The build passed.
<whitequark> cr1901_modern: actually, opengl1 didn't do culling
<whitequark> that was just an even number of bugs
<GitHub> [solvespace] whitequark pushed 1 new commit to master: https://github.com/solvespace/solvespace/commit/683ac78ca25f1e3f6b3468bb81b1cec2ca64cc94
<GitHub> solvespace/master 683ac78 EvilSpirit: Optimize FromTransformationOf for performing identity transformation....
<travis-ci> solvespace/solvespace#212 (master - 683ac78 : EvilSpirit): The build was broken.
<GitHub> [solvespace] whitequark commented on issue #115: @Evil-Spirit Also, this only seems to happen to normals, and not to edges... https://github.com/solvespace/solvespace/issues/115#issuecomment-261557659
<travis-ci> solvespace/solvespace#212 (master - 683ac78 : EvilSpirit): The build was broken.
<GitHub> [solvespace] whitequark pushed 1 new commit to master: https://github.com/solvespace/solvespace/commit/369b0a378cbc7dd58fa1f5485c4c3eadca434764
<GitHub> solvespace/master 369b0a3 whitequark: Optimize toggling visibility of normals, points, edges, outlines....
<travis-ci> solvespace/solvespace#213 (master - 369b0a3 : whitequark): The build is still failing.
<travis-ci> solvespace/solvespace#212 (master - 683ac78 : EvilSpirit): The build passed.
<cr1901_modern> whitequark: I see, that makes more sense to what I remember. Haven't looked at the gl2 code yet
d42 has quit [Ping timeout: 250 seconds]
d42 has joined #solvespace
<GitHub> [solvespace] whitequark commented on issue #87: I've looked into it. We already have a way to select anything unselectable (pick it from the group info), or in most cases you could simply drag an endpoint, and the implementation, and the implementation went ugly very quick, so I'm going to close this.... https://github.com/solvespace/solvespace/issues/87#issuecomment-261585721
<GitHub> [solvespace] whitequark closed issue #87: Clicking "Zero-length segment" message should select segment https://github.com/solvespace/solvespace/issues/87
<GitHub> [solvespace] whitequark commented on issue #9: (SolveSpace has tests now.) https://github.com/solvespace/solvespace/issues/9#issuecomment-261586822
mumptai has joined #solvespace
mumptai has quit [Remote host closed the connection]
cr1901_modern1 has joined #solvespace
cr1901_modern has quit [Ping timeout: 256 seconds]