Re: Patch: small reduction in output from make doc

[Graham Percival]

On Fri, Jun 17, 2011 at 10:28:37AM +0200, David Kastrup wrote:
> "Phil Holmes" <address@hidden> writes:
> 
> > From: "David Kastrup" <address@hidden>
> >> The normal behavior for a Unix utility is to write to stderr when
> >> something goes wrong and then bottom out with a non-zero exit status.
> >> Make (or similar tools) stop, and the output on the screen is relevant
> >> to what command was called last, what went wrong, and where the process
> >> stopped.

Agreed.

> > I would agree that progress messages should go to stdout, and error
> > messages to stderr so that "normal" progress could be distinguished
> > from "important" error.
> 
> The number of things about which people fancy themselves to be in
> agreement with me never ceases to amaze me.  In particular in this
> thread.
> 
> If my communication skills are that abysmal, I should probably just shut
> up.

I believe the problem is this: David uses specific terminology
(i.e. "jargon").  Now, jargon is great when the discussion is
between people who share the same knowledge, but it leads to
confusion when not everybody shares the same specific meaning.

Terms like "make", "non-zero exit status", "stdout", "sterr", and
"BS" have unambiguous technical meanings in the context of unix
build systems.  However, think that the only person in this
discussion who really knows this area is David.
(I don't mean to imply any insult to Reinhold if he's also an
expert)

I personally think that I can *just about* understand what David
is suggesting (of course, there's no guarantee that I'm correct; I
might be over-estimating my understanding).  I _think_ that I
recognize all the jargon which he uses, and I _think_ that I know
the correct meanings of those terms... although I've had to look
up a few things, such as
http://stackoverflow.com/questions/5009258/update-command-line-output

However, I am certain (again, with no insult implied) that Phil
does not know the domain-specific meanings of much of the jargon
that David uses.  This leads to miscommunication.


I think that Phil is doing a fantastic job of learning + fixing
the build system.  Yes, he's not a long-bearded unix guru --
according to his picture on Scafell Pike in 2004, he keeps it
quite close-cropped -- but despite his beard handicap, he's doing
a great job of trying to learn this stuff.

Maybe it would be worth indicating terms with specific meanings?
Something like
  If there is a *non-zero exit status*, *make* will stop and
  display *output* about the *process* which failed
?  then if the reader encounters the word "process" and thinks
"well, I know what the English word means, but I didn't think
there was a special unix definition of that word... I'd better
look it up!" we can hopefully avoid miscommunication.

Cheers,
- Graham