-=> William McBrine wrote to Bo Simonsen <=-

 WM> OK, I've looked at the Maximus code. It's converting long to ieee and
 WM> then ieee to msbin, so it is afflicted by the float format problem.

I take it back -- I see that the existing Maximus *ieee* code uses all
integer math internally, so I'm not sure what the problem is. However, my 
version is still simpler, and simpler is usually better. :-)

On the other hand, if it really is an alignment problem, as you suggested
earlier, then try this patch:


--- mb_qwk.c~   Thu Aug 14 01:30:44 2003
+++ mb_qwk.c    Thu Aug 14 11:28:40 2003
{at}{at} -1509,7 +1509,9 {at}{at}
 {
   if (n_qidx)
   {
-    write(ndxfile, (char *)qidxs, sizeof(struct _qmndx)*n_qidx);
+    int i;
+    for (i = 0; i < n_qidx; i++)
+      write(ndxfile, (char *)(qidxs + i), 5);
     n_qidx=0;
   }
 }


By writing each index individually and hardwiring the size of '5' (could
be changed to a #define for style), it doesn't matter if the _qmndx struct
gets padded for alignment or not. Ideally this would be used with fwrite()
rather than write().

... All that glitters has a high index of refraction.
--- MultiMail/Linux v0.45