Ingar missed this

* Ingar est parti (Ping timeout: 250 seconds)
* StalKermit est parti (Read error: Connection reset by peer)
<illwieckz> perturbed`, thanks a lot you made me able to find the initialization I was looking for
* FreeSlave (~freeslave@46.147.193.183) a rejoint #unvanquished-dev
<illwieckz> I already noticed that the shader code implied that boolean were false by default, for example the `parallax` boolean is only set to true, there is no code to set this boolean as false, so if there was an initialization, it had to be as a whole
<illwieckz> I now found this:
<illwieckz> Com_Memset( &table, 0, sizeof( table ) );
<illwieckz> this is the table of shaders
<illwieckz> so I can assume a boolean is false by default
<illwieckz> `Ishq, it does not yet, it will
<illwieckz> within the boundaries of the current r_dpMaterial PR, the only drawback is extra compute, btw I don't know how the renderer behaves if it autodetect a file suffixed with _norm and then a normalMap keyword Shader set it again with, in the best situation it would only load the image twice
<illwieckz> because we use _n suffix but it prevents no one tu use _norm
<illwieckz> it will have drawbacks and it's better to disable r_dpMaterial by default because normalmap detected the darkplaces way will have a bit set to true telling this normalmap is in directx format
<illwieckz> so if an unvanquished mapper uses names his normal map _norm and dpMaterial is enabled… it may face issues
<illwieckz> he*
<illwieckz> -uses
<illwieckz> so if an unvanquished mapper suffixes his normal map _norm and dpMaterial is enabled… he may face issues*
* Amanieu est parti (Ping timeout: 245 seconds)
* Amanieu (~amanieu@unvanquished/developer/amanieu) a rejoint #unvanquished-dev
* ChanServ donne la parole à Amanieu
<illwieckz> hmmm, I had to delete my glsl cache to get things rendered properly after having switched git branches
<illwieckz> is there ways to better detect the need for a cache invalidation?
* ZdrytchX (~ZdrytchX@unaffiliated/zdrytchx) a rejoint #unvanquished-dev
<perturbed`> another bug with the build system is that errors are not detected when creating the glsl hexdumps
<perturbed`> i know because it "works" when building from a readonly source dir
<illwieckz> another bug with the CI is that we never check for glsl validity at build time
<illwieckz> this is only checked when the engine is run for real
<perturbed`> that's a feature request :P
<illwieckz> perhaps we can run the engine for real in travis
<illwieckz> at least a subset of the engine that compiles the glsl
<illwieckz> it would be cool to have that subset in anyway
<illwieckz> so this tiny engine that only compiles glsl can be called in a make target so we can do make && ./daemon
<illwieckz> daemon would even not be called
<illwieckz> because the test would have been done by make first
* ingar (~ingar@ptr-8a96bawbvr6rowyby1i.18120a2.ip6.access.telenet.be) a rejoint #unvanquished-dev
* ChanServ donne la parole à ingar
* ingar s'appelle maintenant Ingar