Re: seperator error in __mask_snprintf_len

[Paul Jackson <pj@sgi.com>]

A couple of questions on your proposed patch for __mask_snprintf_len()
in lib/mask.c:

 1) Why make the MASK_CHUNKSZ a possible (compile time) variable?
    I can think of a couple good reasons why not to:
    a] So long as we have the current format, in which each word
       is _not_ zero filled, then the chunk size needs to be a
       well known constant, or else the output is ambiguous.
       For example, an output of "1,0" is ambiguous unless we know
       a priori that the "0" stands for exactly 32, say, bits.
    b] Even if we change to a zero filled format, better to just
       always use the same chunk size, as that is one less detail
       to confuse user level code.
    I don't see any reason offhand for needing code that works with
    more than one chunk size.
 2) Why the trailing "buf[len++] = 0"?  Won't the last snprintf do
    as much?
 3) This code has quite a bit more detail of bit shifts, masks and
    arithmetic than before.  Perhaps some is necessary to fix the
    word order bug I had, perhaps some is only needed to allow for
    the chunk size to vary.  I'll take a shot at seeing if I can
    find a less detail-rich expression of this that still gets the
    word order correct.

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373