bug#22241: 25.0.50; etags Ruby parser problems

[Eli Zaretskii <eliz@gnu.org>]

> Cc: 22241@debbugs.gnu.org
> From: Dmitry Gutov <dgutov@yandex.ru>
> Date: Mon, 1 Feb 2016 11:40:46 +0300
> 
> On 01/31/2016 09:11 PM, Eli Zaretskii wrote:
> 
> > Ah, so it _is_ important.
> 
> It kind of is. But I can open a separate bug for it, if you want.
> 
> > But then I'd need a complete specification
> > of what is needed.  (And I already smell a tip of an iceberg.)  Again,
> > the references are scarce and incomplete, but I already understand
> > that it could be either of the following
> >
> >    attr_WHATEVER :foo
> >    SOMETHING ; attr_WHATEVER :foo ; ...
> >    attr_WHATEVER :foo, :bar; ...
> >
> > Is that true?  Are there any other forms, or can the symbol be
> > followed only by a comma, a semi-colon, or whitespece?
> 
> The newline might also be preceded by a comment, I suppose.
> 
> But really, if recognizing attr_WHATEVER when it's just one of the 
> instructions on a line presents a noticeable difficulty, you can 
> disregard that case: nobody really does that in practice. Or we can 
> disregard it at least until somebody complains.
> 
> So you would handle
> 
> attr_WHATEVER :foo, :bar # comment
> 
> and probably
> 
> attr_WHATEVER :bar;
> 
> (the semicolon is redundant, but hey, it shouldn't be too hard to support)
> 
> and the most difficult realistic case I can imagine looks like this:
> 
> attr_WHATEVER :foo, :bar, # comment
>                :qux, :tee

OK, this is all implemented, except...

> > And what ends
> > a line like that -- a newline, or can it be continued on the next
> > line?
> 
> If there's a comma at the end of the current line, the argument list 
> continues on the next one.

...this.  If supporting such split definitions is important, it will
need a slightly more complex code.

> Let's go with my original suggestions, then:
> 
> .rb .ru .rbw Rakefile Thorfile

Also done (and doing so exposed a real bug in etags).

Please take a look.