On 30 Mar 2000, Lars Gullik Bjønnes wrote:
> Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
>
> | >>>>> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
> |
> | Lars> I am thinking more and more seriously about dropping the
> | Lars> lyxstring and use what the system provides instead. Will require
> | Lars> some work, but not a lot. The the sstream problem would just go
> | Lars> away.
> |
> | Yes, we will probably be forced to do that. A consistent STL is
> | prefereable to more complete classes... However, if we do that and are
> | forced by xtl to use exceptions, I fear that code size will explode
> | for no practical result (do we need exception support from xtl?).
>
> Actually the code will not be that much bigger... ok a megabyte or so.
> The unstripped binary will be big. When looking at binary sizes don't
> use ls -l lyx, use size lyx instead that gives the true size of the
> program..
And compile times will increase by at least double. At least that's my
experience so far -- assuming you have plenty of memory and don't start
swapping.
> Not only xtl requires exceptions, conforming omplementations of the
> C++ standard library also requires it.
But most STL implementations can work without exceptions. It should be
fairly straight forward to modify xtl to also compile -fno-exceptions.
It will still require -frtti though no matter what you do to it..
Allan. (ARRae)