Re: new autotest macro: AT_DATA_UNQUOTED

[Akim Demaille]

 >> > Why is it bad to want to do it this way?
 >> It's just a feeling.  Plus, the original author resisted to such a
 >> feature, and I blindly believe in his opinions.  But I'm not making
 >> this a religious matter: if you _need_ this, let's change it.  I
 >> resisted until now because in each case, there was another solution:
 >> - adjusting the wrappers
 >> - introducing macros

 > Fair enough.  I'm still interested in the rationale though.  

That's the question I feared :)  I have no good and sound reason, it's
just something I inherited, and that very mild arguments support:

- each time I saw such a request, m4 was a better answer, or...

- using a wrapper made the test suite cleaner, more robust, and
  with less code duplication

- testing is a delicate art, extremely sensible to variations, and maybe
  pushing the author to write something as invariant as possible helps
  avoiding nondeterminism.  In addition, you have better guarantee to
  have on you desk the very same files as those on the remote machine
  on which the user is fighting against bugs.


So the day a static solution does not exist, we'll move.