README.qtcopy: -no-stl or no -no-stl?
bo at sonofthor.dk
Tue Jan 20 10:30:39 GMT 2004
On Tuesday 20 January 2004 11:05, Andy Fawcett wrote:
> On Tuesday 20 January 2004 11:44, Simon Hausmann wrote:
> > On Tuesday 20 January 2004 10:30, Bo Thorsen wrote:
> > > And another thing: Why the "-no-exceptions"? Ever since gcc 3.0.0
> > > was out, having exceptions support compiled in is free of runtime
> > > costs, up until the point where you actually throw an exception. If
> > > people want to write code that uses exceptions, why not allow them
> > > to do so? I know there's an old grudge against exceptions in kde
> > > code, but there's no reason to cripple peoples Qt, when they might
> > > download Qt apps that need exceptions. And yes, there are examples
> > > of such apps.
> > I believe the original reason for the -no-exceptions recommendation
> > was not runtime cost but the increased size of the generated code and
> > increased memory consumption. Not sure though if that still applies
> > with recent version of gcc.
> We still support some platforms using older gcc versions too (FreeBSD
> 4.x for one, there are probably more). I'm not too familiar with the
> level of exception support on 2.95.x, so maybe this is not so much an
> issue anymore. Any gcc experts care to comment on this?
gcc 2.9x + binutils from the time sucked so much, that we still have
everyone mistakenly believing exceptions are bad, when in fact only the
tool chain implementation was the real issue.
You could add a note saying "If you are using gcc version 2.95.x or 2.96,
you will get a speedup by adding "-no-exceptions to the configure
And before you object too much about this, please remember that only the
combination of a selfcompiled qtcopy + gcc < 3.0 is the issue here. I
don't see why we shouldn't restrict that old workaround to the place
where the actual damage is.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
More information about the kde-core-devel