@lanodan what compiler even supposed to do with unknown directives.

I know they can ignore #pragma and that's what they should do.

@lanodan clang just strips them away lol

I never questioned how gcc preprocesses device trees so I dunno how it works

@a1ba Meanwhile I ignore the hell out of #pragma because there's usually better ways of doing it (like testing if compiler supports a compile flag).

@lanodan the only really useful pragma I know and which is standardized is openmp

some insane people like to put #pragma once (I was there too) but it's useless and doesn't even affect build times.

@lanodan I've even seen some put shit like #ifdef _MSC_VER #pragma once #endif

@a1ba@suya.place Yeah, regular #ifndef _HEADER_H header guards makes more sense.

Or just no header guards, which is typically what I do unless it's a public header you're supposed to #include in an application.

@lanodan I got bitten by HLSDK code where changing the order of includes breaks build so hard, so I tend to make headers compile-able, which in turn increases odds of double-triple includes, so header guards are required.

Or just put everything into one shared header lol that works too.

@a1ba Order of includes is so annoying, in fact it bit me yesterday with Tcl headers…

@lanodan is that a problem of tcl itself? Ouch.

@lanodan so the answer is the -xassembler-with-cpp flag