Re: [Json] Generators

[Marcos Caceres <[e-mail]>]

On Monday, June 10, 2013 at 1:26 PM, Carsten Bormann wrote:

> 
> Comparing two algorithms is much harder (even for humans) than checking an implementation against assertions.
> Even more so if one of them is written in English, so it cannot even be executed unambiguously.

Right, but this is why we have test suites - i.e., to provide unambiguous computationally-verifiable testable assertions of what the English is saying. Personally, the primary motivation for me to see the algorithms is so that we can make sure we have a consistent way of testing everything in the spec. In my experience, the algorithms also help identify the gaps - particularly as you try to test and code against them. Yes, this also means having very competent people writing the tests so to make sure edge cases are also captured.     
> Don't get me wrong: Adding a pseudocode example can do a lot to aid understanding (and to disambiguate poorly worded assertions). However, turning English language pseudocode into the default mode of explaining things is a major mistake (and going full FDT is almost always worse, except for small elements such as ABNF). I'm not going to repeat what can be read as another copy of the cheap shot (and even implicitly blaming a single person for such an outcome is always wrong, mea culpa), but I think you are well aware what might count as exhibit A. Let's not go there.

My experience has been different, I admit. I'm mostly involved with specs at the W3C and we exclusively do algorithmic specs. I won't claim that one approach is better than the other, as I can only speak from my personal experience. I've jotted down some of those thoughts here:

http://www.w3.org/TR/test-methodology/

> Douglas' tight prose in RFC 4627 is actually something I would recommend to emulate.
> The "ambiguities" we are discussing are either the result of very "creative" use or were left in deliberately (SHOULD...).
> That's why I'm a bit unhappy that we seem to have started wordsmithing before getting some of the decisions done.

You are probably right. Like I said in a previous email, I just joined the list and have no idea at what stage you all are at - but I'm happy to try to help. 

-- 
Marcos Caceres