Re: completeness

On 22 Feb 2008, at 10:50, Bijan Parsia wrote:
[snip]
> "Typical non-standard" seems like a "spec smell" to me. If it's  
> typical, let's standardize it. In fact, this is more than typical:  
> It's ubiquitous.
[snip]

I realized after sending this that people might not understand what I  
mean by "spec smell". I mean it by analogy to "code smell":
	http://en.wikipedia.org/wiki/Code_smell

which is, in my understanding, a fairly neutral term (i.e., code  
smells don't necessarily mean there's a problem, just that careful  
examination and justification is required). I think if we find  
ourselves saying, in informative text, that some deviance is "typical  
but nonstandard" we ought to scrutinize very carefully why this  
*typical* behavior is nonstandard. All things considered, it may make  
sense to keep it as "typical but nonstandard", but I think the more  
we can avoid such the better off the spec. (Esp. if implementors end  
up using the informative part to guide the user observable behavior  
of their software.)

Cheers,
Bijan.

Received on Friday, 22 February 2008 12:21:51 UTC