Re: Could --disable-leaks be made a runtime switch?

[Avery Pennarun]

On 12/02/2008, Thomas Dickey <address@hidden> wrote:
> On Tue, 12 Feb 2008, Avery Pennarun wrote:
> > On 12/02/2008, Thomas Dickey <address@hidden> wrote:
> >> I recall seeing SuSE's package for ncurses had turned on the feature -
> >> a while back.
> >
> > Perhaps it should be the default nowadays?  The tiny performance
> > increase from not cleaning up the objects should be irrelevant on
> > modern machines, while the valgrind improvement will save people time.
>
> most of the cleanup is a one-way process that's done just before exit.

Right... and --disable-leaks makes this cleanup happen.  What's the
rationale for not doing the cleanup process?  Apparently SuSE prefers
to include it, so it can't be anything fatal.

Thanks,

Avery