alignment alternative

[David Relson]

At 07:18 AM 10/29/02, Gyepi SAM wrote:

>On Tue, Oct 29, 2002 at 12:11:31AM -0500, Clint Adams wrote:
> > This determines the alignment boundary at configure-time, rather than
> > compile-time.  It uses a method that autoconf upstream calls "sadistic",
> > and provides no real advantage over the other method.  Nevertheless, it
> > seems preferable to me.
>
>You are using the return value of the test program as the alignment 
>requirement?
>That does seem strange, but would work since the value is less than 255.
>
>Alternately, we could have the test program print out a #define into a 
>header file that we
>include.
>
>I was quite surprised to discover that the autoconf does not have the 
>facilities to read
>the output of test programs.
>
>-Gyepi

Gyepi and Clint,

Figuring out the alignment with configure is good.  Using the exit code is 
not.

I like the idea of adding a #define to a header file.

We currently have globals.h and common.h.  I suggest that we have a 
globals.h.in with a "#define BOGOFILTER_ALIGNMENT value" in it.  Then use 
sed to generate globals.h from it, with the correct value.

David