Re: Prefer error-verbose to error_verbose

[Joel E. Denny]

On Fri, 17 Apr 2009, Akim Demaille wrote:

> Interesting question.  For the union, currently dots are allowed and not
> reported by Bison, leaving this task to the compiler.
> 
> %token accepts periods, but only C-identifiers are output in the parser file.
> 
> address@hidden /tmp $ cat foo.y
> %token foo.bar
> %token foo_bar
> %%
> exp: foo_bar foo.bar;
> address@hidden /tmp $ grep foo.bar foo.tab.c
>      foo_bar = 259
>   "$end", "error", "$undefined", "foo.bar", "foo_bar", "$accept", "exp", 0
> 
> 
> We could continue like this.

I can't see any reason why dashes should be worse than periods in these 
cases, so that makes sense to me.

> So do you agree with this patch?

I like it.

> (Is there a significant difference between "dot" and "period"?  Internet
> search is not really helping here.)

In my opinion (but I don't know how to validate it), "period" 
unambiguously refers to the "." character while "dot" could mean any 
dot-like character.  Of course, I don't see any other dot-like characters 
on my keyboard, but there's always unicode.

On the other hand, I think of a period as a punctuation mark at the end of 
a sentence, and that's not how we're using it here.  "dot" is more 
general.

The phrase "period character" seems to be unambiguous and avoids 
suggesting any semantics, so maybe that's best for formal documentation.