FOSS gamedev and creative worlds

Ok so this looks awesome BUT my worlds wont load,

Here is the debug Text

-------------
Separator
-------------

In trans_func.
In trans_func.
12:35:47: ERROR[main]: Some exception: "Illegal instruction"
12:36:14: INFO: event_handler(): Ctrl+C, Close Event, Logoff Event or Shutdown Event, shutting down.

post this in bugs and problems

well i cant so too bad

Cooper wrote:well i cant so too bad

Why not?

From what I hear, it means your operating system or your processor or something is out of date. I forget which.

If you're using mods, make sure you've updated them to their latest version. Very old versions of More Ores can cause divisions by zero (which can be the cause of an "In trans_func." error).

Illegal Instruction refers to a PROCESSOR THAT DOES NOT HAVE SSE2, which is required by Minetest 0.4.x

Also: You have a Pentium 3 or AMD Duron. So. You can't play.

Jordach wrote:Illegal Instruction refers to a PROCESSOR THAT DOES NOT HAVE SSE2, which is required by Minetest 0.4.x

Nope, it isn't.

xyz, most often, it is

Jordach wrote:xyz, most often, it is

Minetest itself doesn't require any specific processor features.

Yes it does. Try 0.3.1 on a Intel Pentium 3; or AMD Duron. It needs SSE2.

Minetest itself doesn't require any specific processor features.

Yes it does

i think is more compile time related thing. at least gcc have -msse2 and -mno-sse2 as example.

Jordach wrote:Yes it does. Try 0.3.1 on a Intel Pentium 3; or AMD Duron. It needs SSE2.

I don't have any of them but I know some people were able to run it. (and, well, I won't bother arguing with you on that anymore)

CryAngel wrote:

Minetest itself doesn't require any specific processor features.

Yes it does

i think is more compile time related thing. at least gcc have -msse2 and -mno-sse2 as example.

Yes, it only depends on build.

I was going to try making a build with -mno-sse2, but when I got to trying it I realized that one would probably have to find ways to re-build all of the dependencies with the flag as well and, due to the nature of Windows building, they all have substantially different build systems. Then, even if one thinks that one's succeeded in re-building without SSE2, one wouldn't know for sure until Cooper or someone else with an outdated processor tried it; thus, they would have to have not been scared off by Jordach.

It might be more of a question of who has enough guts to do that much work to make something that he cannot test directly and doesn't even know for sure will solve the problem.

Then again, I'd like to be proven wrong; I've had more than my fair share of difficulty building stuff on Windows…

If you use the crosscompile buildbot, you can build all dependencies the same way.