Did you look at the script on SourceForge which builds the ./configure line
automatically?

As to configure, well, the cost of correcting every possible error and
anticipating every stupid thing people could do was well over 1000 lines of
fragile code.  So we decided to let people run off the road and hurt
themselves.

It's the difference between driving in the US and in New Zealand (at least
when we were there in '85).  The US is so over protective, that a 2' drop
has a guard rail.  As a result, people EXPECT them and drive accordingly.
In NZ, if there's a guardrail, it really, really meant something and so it
was respected.

Don't believe me?   Look into the research on Anti Lock Breaking systems
(ABS) - it shows very little difference in safety...  As soon as people got
used to what ABS did for them, they increased their average speed, depending
on ABS to 'save' them.

I suppose I could add the tests for the redundant leg in each triangle and
warn the user... yeah, that's about 6 lines + msg in the .m4 file...

As to why it failed, I still believe it was the myrrd 'stuff' conflicting
with the installed .so, but somebody won't get rid of it, so I've washed my
hands...  I don't even bother responding if it's a myrrd problem.
Personally? I just delete that piece of the source tree and/or remove it
subdirectory from the Makefile when I build working copies of ntop's
development version.


-----Burton


-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of The
Jetman
Sent: Wednesday, May 28, 2003 4:44 PM
To: [EMAIL PROTECTED]
Subject: Re: [Ntop] rrd plugin doesn't load when ntop starts


----- Original Message -----
From: "Burton M. Strauss III" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Wednesday, May 28, 2003 15:29 PM
Subject: RE: [Ntop] rrd plugin doesn't load when ntop starts


>   When I CONFIGUREd NTOP w/ unnec (for BSD) --with-gdchart
>
> Understand that while they may not be necessary (IF the .so libraries
> exist), the presence or absence of these configuration parameters causes
> ntop's ./configure to perform different tests.
>
> We used to hand hold you through this, but that causes 1000s of lines of
> configure.in code.
>
> Instead, silly ntop, we belive and use what you tell us.
>
> If you drive a car off a cliff, bad things happen.  If you tell a program
to
> configure itself stupidly, bad things happen.
>
> The only difference, between 2.1 and 2.2 is that we removed the
guardrail...
>
>
> -----Burton

    Burton:  I understand all of that, but fact is, I spent a couple of
hours tracking that particular kind of error down (after more than a dozen
configure/make cycles) and what ultimately resolved it was to simply use
ONLY those essential options (for FreeBSD), bec somehow CONFIGURE was
getting confused.  (As you well know, it's a non-trivial script.)  This
particular box failed only at the point of rrdPlugin and I got NTOP (w/
working RRD support) the instant I made this change.  That's the reason I
now have my own little script for automagically and reliably bilding NTOP.
It's never failed (under BSD.)

    One of these days, I'll probably be inclined to bring up NTOP on a
Linux box and I'm CERTAIN the quirks to make it work will be different,
but in the end, NTOP WILL work....Jet

===============  From the desk of Jethro Wright, III  ================
+  Never attribute to malice that which is adequately explained by   +
+  incompetence.                                                     +
===  [EMAIL PROTECTED]  =====================  Hanlon's Razor  ===

_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop

_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to