I'm looking at the C source to the ZLIB compression library and have
run into this line:

Call UPDATE_HASH() MIN_MATCH-3 more times

The Q is what the heck does "Call" mean?

I have a vague recollection that's it's a preprocessor directive and
it certainly looks and feels like one (UPDATE_HASH is a macro) but I
can't find any documentation on. (Try doing a keyword search on "call"
and see how many pages you get. :)

Anyone provide a link of the syntax?

Looks like the place to start to me.
I see links there to a FAQ and a manual.

and please restrict your posts here
to the topic described therein.

Thank you.

-Larry Jones

I'd just never actually ran into code where the author purposely made
a conditional where it would crash the compiler with plain english
before. What was confusing me is that there actually WAS a variable
named "more" defined a few lines before the statement, etc.. Was
thinking it was all some high black-magic C that no one ever let me in
on. ;-)

This is on-topic, I think, but only by the slimmest of margins.  It's
really a question about reading C source code.

You didn't include a sufficiently large fragment.  This line appears
twice in deflate.c, and in context reads:

#if MIN_MATCH != 3
             Call UPDATE_HASH() MIN_MATCH-3 more times
#endif

which is a clue.  When you compile this source, what is MIN_MATCH
defined as?  It's 3, and this line is removed during the preprocessing step.

This line is not C.  It's English, and it's a comment directed to
maintainers.  It's not a C comment (ie surrounded by /* */) presumably
because the author (Jean-loup Gailly, apparently) felt it's sufficiently
important to cause an error during translation (which it would, since
the UPDATE_HASH macro takes three parameters and they're not supplied
here, which is a constraint violation and requies a diagnostic, even if
there weren't other problems).  The idea is that if some clever sort
mucks with the definition of MIN_MATCH, they'll get an error from the
compiler which will prod them into understanding how deflate.c works and
what else they need to change to hope to have a working program.

There are better ways to do this sort of thing, of course, notably the
#error preprocessing directive.  But perhaps the author wasn't familiar
with #error, or wanted to support some broken implementation that didn't
support it.

The m{*filter*}of the story: Preprocessing directives are part of the source,
and if you want to understand the source, you'd best understand the
preprocessing directives.

The problem was there is a LOT of macro expansion code going on in the
source code (as you are probably well aware at since you've since the
C source). There was also a variable called "more" declared a few
lines few before the line in question.

In other words, except for "times" everything else was an in-scope
macro or variable. After a very long day ripping apart obscure macro's
(I'm tasked with converting the entire thing to standard C++) ...
well... just needed to step back and get some fresh eyes . :)

Just a classic case of over-thinking a problem.

I think that's ISO standard across platforms now though isn't it?
Maybe not. I'll have to look that up later. ... Ah. It does seem to be
part of the ANSI standard. Specifically:

"The directive #error has been introduced to provide an explicit
mechanism for forcing translation to fail under certain conditions.
...  Traditionally such failure has had to be forced by inserting text
so ill-formed that the translator gagged on it. "

Cool.

I voluntered to go over the 1.2 beta's of zlib. Adding a nice #error
would have to be at the top the "fix me" list I'd have to say. ;-)