J3/16-161
To:         J3 Members
From:       Van Snyder & Bill Long
Subject:    Binding label names with zero length
Date: 2016 February 08
References: 16-007, 16-142

Discussion
----------

Paper 16-142 included these two questions:

[483:17 15.9.2p2] What is the binding label if NAME= appears but the
string has zero length after trimming leading and trailing blanks?

[484:27 15.10.2p2] What is the binding label if NAME= appears but the
string has zero length after trimming leading and trailing blanks?



Relevant text in 5.5.5 constrains the NAME= value in a
<language-binding-spec>:

"If the value of the <scalar-default-char-constant-expr> after
discarding leading and trailing blanks has nonzero length, it shall be
valid as an identifier on the companion processor."

This is a limitation on the syntax. It appears to allow the
possibility of len=0. But that certainly cannot be the identifier on
the companion processor, and the name is supposed to such an
identifier.

Options are:

  1) zero length is actually not allowed, which should be specified in
  the standard. In this case, a zero-length name is a programmer error.

  2) zero length is the same as NAME= not being specified. Probably
  don't want this one. Implementation issues, and unexpected behavior
  for users.

  3) Do nothing.


Edit to 16-007:
---------------

Option 1:

[94:9-10] In 5.5.5 BIND attribute for data entities, replace paragraph
2 by:

"The value of the <scalar-default-char-constant-expr> after discarding
leading and trailing blanks shall have nonzero length and it shall be
valid as an identifier on the companion processor."

Option 2:

TDB

Option 3:

(none)