Re: TMCL Regexp constraint

Am 01.12.2010 um 21:18 schrieb Lars Marius Garshol:

> * Robert Cerny
>>
>> Thanks for leading the way to the well hidden answer. I agree that  
>> it should be that way.
>
> Lars Heuer is right, of course: the correct answer is:
>   a. the boundaries are implicit
>
> That is, if your regular expression is "foo", it's interpreted to  
> mean "^foo$".
>
>> But i must admit the example threw me off and would suggest a  
>> change so that it becomes clearer. The current example matches any  
>> string which *contains* a phone number. I think a better example  
>> would be any string that *is* a phone number.
>
> The example requires the value to actually *be* a phone number,  
> just like you suggest. I guess I mislead you when I mistakenly  
> wrote that (b) was the answer.

It is more likely that my confusion was caused by wishful thinking :)

While this convention obviously makes sense - now that i know it, i  
would not want to live without it (until i forget it) - it also  
increases the requirement for the technology user to *know* things  
and thus learn, in this case relearn (a tiny little bit) the notation  
of regular expressions. I picture the whole situation as a medieval  
castle, the technology newbie needs to get in the castle and the spec  
is the fortification. Once he makes it in he is very happy and feels  
this strong bond with all the others that made it (shoulder tapping  
happens): after all convention unites as much as it separates. What  
nobody told the guy is that it is much more difficult to leave the  
castle. Anyway he probably  would not want to go since he invested so  
much to get in.

So what advice can the SC34WG3 draw from this simple undebatable truth:
* create bigger specs that are harder to understand
* do not rely only on pitfalls inherited from other standards, create  
your own pitfalls
* every once in a while throw in a sentence that totally confuses  
everybody, it adds another day to training
* get more people in the boat, preferably the ones that are convinced  
that things should go their way

.-)

Robert