Re: monit CHANGES.txt gc.c l.l monitor.h p.y util.c...

[Michael Shigorin]

On Mon, Dec 20, 2004 at 11:20:19AM +0100, Jan-Henrik Haukeland wrote:
> >BTW, is it reasonable to put at the back of the head something
> >like "make separate parser which is called when monit starts
> >or reloads and doesn't sit in memory with all the libraries
> >involved full time"?
> I think you mix-up what parsing and a parser-generator is with
> something else.

No I don't.  It's just a parser (the result being linked in a
running monit) ends up being one-time chunk of code.  It's like
root partition, when it's small, it's harder for a stray bad
block to render the whole system unbootable to the minimal state.

When the process image is smaller, it's less prone to being
swapped out; when its share library deps are tighter, it's less
prone to ABI breakage and .so corruption, etc.

Sorry for being *sooo* obnoxious :-)

> The parser-generator (YACC) and scanner (LEX) generate a
> state-machine in plain C code which is compiled and linked into
> the monit executable as any other C code.

I've played with yacc/bison for some time :-)

> Don't confuse C with a high-level language like Ruby :-)

Sure I don't! :-) (...and thus not writing system-level code)

-- 
 ---- WBR, Michael Shigorin <address@hidden>
  ------ Linux.Kiev http://www.linux.kiev.ua/