[m-dev.] for review: fixing formatting of HLDS dumps

[Fergus Henderson]

On 19-Feb-1999, Peter Schachte <schachte at cs.mu.OZ.AU> wrote:
> On Thu, Feb 18, 1999 at 07:47:53PM +1100, Fergus Henderson wrote:
> > It would be better, IMHO, to use an error message
> > like this:
> > 
> > 	...
> > 	foo.m:001:   error: undefined symbol `\n /0'.
> > 
> > (Trying to print out the symbol name in quotes, i.e.
> > 
> > 	foo.m:001:   error: undefined symbol `'\n '/0'.
> > 
> > would be worse, IMHO.)
> 
> I think either
> 
> 	foo.m:001:   error: undefined symbol '\n '/0.
> 
> or
> 
> 	foo.m:001:   error: undefined symbol `\n '/0.
> 
> would be much better (but preferably the first).

In this particular case, I agree.  But you have to look
at the global picture.  Error messages like this one
are *very* infrequent, so it's worthwhile to do
worse on this error message if it means that we can
do a little bit better on most other error messages.

> And for consistency
> 
> 	foo.m:001:   error: undefined symbol foo/0.
> 
> seems perfectly understandable.

That is what we used initially, but our experience was
that using `...' made the error messages easier to read.

> Alternatively
> 
> 	foo.m:001:   error: undefined symbol: foo/0.
> 
> might be a little clearer.

That doesn't work so well for error messages where there
are more than one symbol that we need to name.

> In any case
> 
> 	foo.m:001:   error: undefined symbol `\n /0'.
> 
> seems pretty confusing.  What will the user think when the error is a
> little different and she reads:
> 
> 	foo.m:001:   error: undefined symbol `\n /0/0'.
> 
> ?

That example is not likely to occur in real life.

A much more realistic example is this one:

 	foo.m:001:   error: undefined symbol `//2'.

The ideal thing to do would be to use bold type or a special
colour, rather than using `...'.  That would avoid abominations
like

 	foo.m:001:   error: undefined symbol `'/'/2'.

while still keeping the appropriate emphasis in the usual case. 
Unfortunately, however, there's no portable way of doing that.

-- 
Fergus Henderson <fjh at cs.mu.oz.au>  |  "Binaries may die
WWW: <http://www.cs.mu.oz.au/~fjh>  |   but source code lives forever"
PGP: finger fjh at 128.250.37.3        |     -- leaked Microsoft memo.