On 5/26/05, Eric Dorland <[EMAIL PROTECTED]> wrote: > * Oleg Verych ([EMAIL PROTECTED]) wrote: > > Strange... Didn't notice new gmail's feature with saving drafts, sorry. > > > > Patches also are in there: > > <http://flower.upol.cz/~olecom/bugs/m-264912_fpu/patches/> > > Ok, what does this patch do exactly? Is there an upstream bug with > this patch? >
In short. glibc's libm reported by its developers to be worked only in intel's extended double precision mode, while ECMA Script's math must work in IEEE 754 double precision one. This patch sets FPU on i386 and x86-64 Linux(glibc) in this mode. In mozilla JS engine on Linux uses old fdlibm codebase <http://www.netlib.org/fdlibm/> for all math (see js/src/jslibmath.h) which requires it (see <http://www.netlib.org/fdlibm/readme>). Also i've cleaned Makefiles for this. See <https://bugzilla.mozilla.org/show_bug.cgi?id=264912> for more. And last. In funcion i've put initialization code i saw goto, i couldn't be patient <http://www.acm.org/classics/oct95/>. -- olecom
