Formal review of dtoh

[Dicebot]

On Thursday, 27 March 2014 at 18:28:28 UTC, Dicebot wrote:
> On Thursday, 27 March 2014 at 13:25:14 UTC, Adam D. Ruppe wrote:
>> That's what I was hoping it could be pulled, with the note 
>> that it is super experimental, so maybe people will try to use 
>> it and file some bugs as to what really sucks. The whole 
>> approach might need to be abandoned in favor of a new dmd 
>> switch or something too. That's also why the style doesn't 
>> match and stuff like that; if it needs a rewrite anyway, no 
>> point getting worked up over brace placement. (It did need to 
>> be mostly rewritten a few months ago because dmd changed the 
>> json output!)
>
> You still should provide better CLI, it does not rely on 
> internal implementation and will make easier for random 
> reviewers to test / experiment.

Want to extend this point a bit more : I believe that tools 
contributions should be evaluated based on a principle similar to 
Phobos modules - implementation may have flaws (or even changed 
completely) but user API (CLI + behavior for tools) should remain 
solid once merged. So it needs better attention even if you are 
going to re-write rest of the program later.

I'd consider it a blocker if this to undergo formal voting.