const debacle

[Bruno Medeiros]

Janice Caron wrote:
> I think I've got it.
> 
...
> 
> I honestly think this might be problem solved. But please, please,
> please chime in if I've got anything wrong.

Well, *DUH*!
Yes, your proposal looks safe and sound, and you managed to follow the 
whole logical progression to reach this conclusion. But all of it was a 
trail that had been walked before! Let me explain.

A mechanism exactly like yours has already been described in an academic 
paper called Javari, about adding a transitive const system to Java 
(http://groups.csail.mit.edu/pag/pubs/tschantz-refimmut-mengthesis.pdf). 
The mechanism there is called "romaybe" (short for read-only maybe), and 
is pretty much the same as yours. Quotes:
  "romaybe, a syntactic convenience that reduces code duplication;"
  "Javari provides the keyword romaybe to declare that a method should 
be templated over the mutability of one or more formal parameters—that 
is, the method should have different signatures in the mutable and 
read-only types."


Nearly a year ago, someone posted the link to the Javari paper. A few 
months later, when the const discussion was starting up and gaining 
prominence, I mentioned, quote: "I would say reading it [the Javari 
paper] should likely be  mandatory for anyone who wants to venture 
thinking and posting how the const system should work." 
(http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D.learn&article_id=3880)
That's because the paper and the proposal genuinely look to be very well 
written and thought out, well detailed, and having all the issues 
considered. (unlike this other proposal: 
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4211070).

Hence my comments like, "I wish I had an euro every time someone posted 
about something that had already been discussed and concluded in Javari."

Now, you weren't yet in the D NG when Javari was mentioned, so I'm not 
disparaging you about all of this. In fact, it's a good sign that you 
reached the same conclusion as the one mentioned in the paper.
What is really annoying me, is that Walter (and Andrei), the masters 
behind D's language design, seem to be utterly ignoring such a good 
repository of knowledge and technical wisdom. They seem to be designing 
const with more of a gun-ho attitude, without carefully considering all 
aspects. That's why we had like 3 or 4 versions of const so far, and 
frankly, unlike Walter, I don't think the current system has "mostly 
been worked out". The 'inout'/'romaybe' is just one example of a kink, 
but are many other aspects at fault. (more on that on another thread).


(BTW: the Javari paper also mentions logical const, and offers a 
proposal very similar to your's 'unpaintable')

-- 
Bruno Medeiros - Software Developer, MSc. in CS/E graduate
http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D