.
Last update: 1997-05-20
9945-2-45
_____________________________________________________________________________
Topic: pmatch rules
Relevant Sections: B.5.2
Classification: defect
Defect Report:
-----------------------
(from Andrew Hume Doug McIlroy)
Issue F
________________________________________
[17] In B5.2, the meaning of lines 356-361 is unclear.
Proposed Solution:
Replace the five rules with:
(1) The nesting of substrings denoted by nonnegative off-
sets in pmatch shall be the same as the nesting of
subexpressions in pattern.
(2) A substring denoted by nonnegative offsets in pmatch[i]
shall be the last of all matches to subexpression i.
(3) The offsets in pmatch[i] shall be nonnegative if possi-
ble according to rules (1) and (2), otherwise the off-
sets in pmatch[i] shall be -1.
Rationale:
One of us understood, ``the match ... reported in
pmatch[i] shall be ... within the substring ... rather than
the whole string,'' to mean that the offsets in pmatch[i]
were to be measured relative to an origin specified by
pmatch[j]. We now believe that the rule merely says that
reported matches must nest the same way as the subexpres-
sions. The proposed solution captures this idea succinctly.
Rule (5) could be kept, but nobody needs it, given lines
289-90.
WG15 response for 9945-2:1993
-----------------------------------
The standard states the required behavior and
conforming implementations shall conform to this.
Concerns about the wording of this part of the standard have
been forwarded to the sponsor.
Rationale for Interpretation:
-----------------------------
Interpretations cannot make substantive changes to the standard.
This may be considered for a future revision.
_____________________________________________________________________________