Iain Buclaw Wrote:
> Anyway, yes I believe this would be the expected behaviour. GDC has a slightly
> different strategy than DMD to ensure symbols that are supposed to be only
> linked
> once are done so correctly.
Haha, just by making sure they're always linked? :)
What is this strategy? I me
Hi there,
I'm trying to follow the instructions at
https://bitbucket.org/goshawk/gdc/wiki/Home to build GDC from the mercurial
repos, but I've not been successful. Not being familiar with the GCC build
process myself, I was wondering if anyone could shed a bit of light on the
problem?
I shoul
By disabling the GC, I've managed to get GDC built on ARM (Linux), and
using the patches on my ticket in Github, I've gotten Phobos and
Druntime to compile, but I'm running into issues when I try to use
fibers. Here's a simple test program:
--
import core.thread;
import std.stdio;
void mai
== Quote from Kagamin (s...@here.lot)'s article
> ---
> struct BL{int OAT=42;}
> BL[1024][64][4] bloat;
> void foo()
> {
> bloat[0][0][0].OAT=1;
> }
> void goo(){}
> ---
> I compile this file and archive it to a library and then link to another file.
> If goo is used, bloat also gets into exe
---
struct BL{int OAT=42;}
BL[1024][64][4] bloat;
void foo()
{
bloat[0][0][0].OAT=1;
}
void goo(){}
---
I compile this file and archive it to a library and then link to another file.
If goo is used, bloat also gets into executable, which is not true for dmd.
Is this expected behavior or
