On 30/01/2009, at 8:37 AM, Mittelbach, Frank wrote:

> I wonder if we might want the concept of _unsafe functions, ie those  
> that do not check but expect the programer to take care of that,  
> while by default all others will be safe. In certain applications I  
> could see speed/processing reasons for something like this. On the  
> other hand one can question how much this matters these days.
> opinions?

I wouldn't be unhappy to see these sort of _unsafe functions, but I  
would wonder how often they'd be necessary (especially in this case  
since padding an argument with quarks and testing if the retrieved  
head is \q_no_value or whatever neither breaks expandability nor takes  
very long...right?).

Let's keep it in mind when we change the tlist functions.