This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

bounds-check mixed with unbounded (was: Re: bounds-check intentionallyundocumented?)


(When replying to gkm, I mucked up gcc@gcc.gnu.org on the
CC-line, then found out that 193.13.178.2 is in ORBS -again.
And nobody likes my patches.  ;-)

> From: Greg McGary <gkm@eng.ascend.com>
> Date: 06 Jan 2000 10:57:24 -1000

> The new BP feature set offers two ways to mix BP and non-BP code:
> 

[ enumerating how to specify unbounded library functions (1 .. 2) ]

Maybe you thought of these and they're bad for some reason, but anyway:

 (3) Have pragmas, (preferrably the new _Pragma () rather than
     #pragma) to make un/boundedness applicable to all
     declarations after.  Better than "__unbounded {}" IMHO.
      Something like _Pragma (unbounded) or _Pragma (bounded),
     that changes a default based on -fbounds-check.

 (4) For multiple declarations, merge explicit or pragmaized
     un/boundedness into those declarations where it isn't
     specified.  (Not sure how feasible this is).

 (5) Have whatever "__unbounded {}" or "_Pragma (unbounded)" be
     automatically applied (under control) to system headers,
     in a manner similar to how implicit 'extern "C" {}' is
     generated.
      See NO_IMPLICIT_EXTERN_C and the applicable field of
     INCLUDE_DEFAULTS.

Thanks for working on this bounds-check thingy.

brgds, H-P


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]