Re: [MinGW-dvlpr] default target architecture?

[Dimitri Papadopoulos-O. <dim...@ce...>]

Hi,

> The reports are really too vague to say what the problem is.  If you
> have a testcase that shows an arch/tune bug, please submit a bug report.

They're not that vague, at least the Qt problem has been reproduced by 
by at least 4 users while building Qt 4.1.1 or Qt 4.1.2 and fixed by 
passing -march=i586 as an argument. That's a test case in itself - 
although very far from minimal

What additional information would they need to provide to submit a bug 
report? These users are probably not able to strip down Qt to a minimal 
example that reproduces the bug and I don't have a computer old enough 
to reproduce the problem myself.


> It might be reasonable to change default arch to 486, but I don't know
> what that will buy.  I would be against changing default arch to i586.
> Some folk (myself for one) still expect to be able to execute
> mingw-compiled programs on a 486.  Believe-it-or-not, some people still
> use 486's for real world apps.

Actually it's the other way round. It looks like MinGW generates code 
that doesn't run on old Pentiums and that -march=i586 fixes the problem. 
So I thought MinGW generates instructions that run only on very recent 
processors, maybe i686. If that's not the case, then this is not an issue.

Dimitri