slipher

> > Map entities can be given names in the entity lump (using name/targetname/targetname2/alias keys) > > These are not usable because they already have semantic meaning to the gamelogic....

> > I don't see why these can't be used. An entity can have up to 4 names so you should be able to fit a new, unique name even...

When the bot gets knocked off its planned course, most of the time the recalculation is triggered by this `PointInPoly` test: https://github.com/Unvanquished/Unvanquished/blob/3a0886e1ba5389e306dcf520ef5a920a31d13849/src/sgame/botlib/bot_nav.cpp#L304-L309 When following a navcon, probably we should have...

Well, it wasn't written as a logic error. The diff adding it: ``` - set_c_cxx_flag("-D_FORTIFY_SOURCE=2" RELEASE) - set_c_cxx_flag("-D_FORTIFY_SOURCE=2" RELWITHDEBINFO) - set_c_cxx_flag("-D_FORTIFY_SOURCE=2" MINSIZEREL) + if (USE_HARDENING OR NOT MINGW) + #...

There are other flags enabled only with USE_HARDENING.

0.55.2 doesn't even use the same caching scheme, so it probably wouldn't cause a bug. We need to test different branches since the recent caching rewrite

I got this with Nvidia GTX 1070 on Windows. I don't know what the cvars were or how to reproduce it.

That's a good question. I never thought of the possibility that the program binary could be different due to changing which GLSL extensions are enabled.

Besides the possibility of the compiler backend outputting different stuff based on extensions, as I was imagining in the previous comment, there's also an easy way to break it. We...

I mean for the stuff like ``` #if __VERSION__ > 120 out vec4 outputColor; #else #define outputColor gl_FragColor #endif ``` We can get two different preprocessed source codes depending on...