On Thu, Mar 11, 1999 at 09:19:27AM +0200, Jukka Marin wrote:
> 
> Well, how about adding the names in the tag system I suggested.. the name
> length would also be specified as one field, so the name could easily be
> skipped.  Actually, it might be faster to have a total length of one entry
> in one field so the entry could be easily skipped..

  My only concern here (so far) is that most data values are
likely to be 4 bytes or 8 bytes.  The tag names are likely to
be on the order of a dozen bytes or more (mem_used_bytes is
14).  By including the tags, we could easily be doubling or
tripling the amount of information being passed from the kernel
to userland, even for applications that don't need any of that
goo.  The verbose version would probably end up being a couple
of K, so it's not so small that size is insignificant.

  If this is not considered an issue, and we don't mind adding
another binary to /usr/bin, then I think your idea would work
great.

  Brian