>> 2. It clutters the code without adding much information.
> It tells lint that you knowingly ignore the value because you don't
> care.

Which is more important, readability by humans or low levels of
warnings from lint?

> Well, show me a compiler that I can use in NetBSD that can do what
> lint does: [...]

I think it's not so much a question of matching lint's capabilities as
one of doubting whether lint is of significant value.  It certainly is
for me; I've tried to use lint, and even with a wrapper to quell the
reams of noise ("pointer casts may be troublesome", furrfu!), it still
is not worth wading through the output of.

> 2. [Compilers] only produce warnings for a single compilation unit.
>    They don't check for consistency between compilation units.

Actually, gcc does much of this with a little help -
-Wstrict-prototypes -Wmissing-prototypes, plus the programmer
discipline to make them useful by not just throwing declarations into
.c files.

> 3. They don't warn about implicit promotions, or if assignments that
>    lose accuracy because of narrowing of types.

Except that I, at least, find that such things happen so often without
being problematic that it's just not worth wading through such warnings
against the chance of finding one that's actually useful - and noticing
it amid the ones that aren't.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B