Re: ANNOUNCE: GNAT 3.15p release

[Simon Wright <simon@pushface.org>]

"J�r�me Haguet" <jeje66@despammed.com> writes:

> We faced a small problem with gnat 3.15p and -s switch for gnatmake.
> For some compiler switches combinations, it does not recongnize that this is
> the same combination.
> See example above.
> This occurs with linux and windows versions.
> 
> Did I miss some changes in this new version ?
> Did I miss something in the documentation ?

I know this problem occurs on 3.15a1 built for Linux from sources, but
thought it was fixed for 3.15a1 binary for Windows. I believe it's
fixed for Linux in a 3.16 wavefront, but that won't help 3.15p users!

(the Linux/Windows business above may be down to my misinterpretation
of an ACT remark and the symptoms, of course. I did wonder why there
would be a difference ..)

For what it's worth, the problem also exists in GCC 3.3 -- at least if
you use any optimisation.

   smaug.pushface.org[19]$ gnatmake -c -u -s -g user.adb -gnatwaL -f
   gcc -c -g -gnatwaL user.adb
   user.adb:10:18: warning: condition is always False
   user.adb:17:16: warning: condition is always False
   smaug.pushface.org[20]$ gnatmake -c -u -s -g user.adb -gnatwaL   
   gnatmake: object up to date.
   smaug.pushface.org[21]$ gnatmake -c -u -s -g user.adb -gnatwaL -O2
   gcc -c -g -gnatwaL -O2 user.adb
   user.adb:10:18: warning: condition is always False
   user.adb:17:16: warning: condition is always False
   smaug.pushface.org[22]$ gnatmake -c -u -s -g user.adb -gnatwaL -O2
   gcc -c -g -gnatwaL -O2 user.adb
   user.adb:10:18: warning: condition is always False
   user.adb:17:16: warning: condition is always False

You see that [20] correctly realised there was nothing to do.