> Any comments on applying this? It looks good to me.
It's amazing that with all of the testing this is the first indication
of a problem (with code which has been through several revs even).
Have you tried the patches through the regression tests?
- Tom
> > Your name : Diab Jerius
> > Your email address : djerius@cfa.harvard.edu
> > Architecture (example: Intel Pentium) : Sparc
> > Operating System (example: Linux 2.0.26 ELF) : Solaris 2.5.1
> > PostgreSQL version (example: PostgreSQL-6.1) : PostgreSQL-6.1
> > Compiler used (example: gcc 2.7.2) : SunOS cc 4.0.1
> > Various compile errors concerning overflow due to shifts (<<) and
> > masking of chars with constants. The first set of errors is due to
> > the compiler holding the result in a signed int ( e.g. 15 << 27 is
> > too big for a signed int, bug fits just fine in an unsigned one)
> > while the destination was unsigned.
> >
> > The second was caused by masking a signed char with a number greater
> > than 127; this caused a warning about the result not fitting.
> >
> > There were various spots where old K&R type prototypes were present
> > and overrode true prototypes specified elsewhere.
> >
> > the sparc_solaris-cc file should have more strict ANSI flags.
------------------------------
