|[5 earlier articles]|
|Re: Command Line Interpreter email@example.com (Steven D. Majewski) (2000-07-29)|
|Re: Command Line Interpreter firstname.lastname@example.org (2000-07-31)|
|Re: Command Line Interpreter email@example.com (Gregory Bond) (2000-08-04)|
|Re: Command Line Interpreter firstname.lastname@example.org (Chris F Clark) (2000-08-04)|
|Re: Command Line Interpreter email@example.com (Scott Nicol) (2000-08-04)|
|Re: Command Line Interpreter firstname.lastname@example.org (Frithiof Jensen) (2000-08-04)|
|Re: Command Line Interpreter email@example.com (Tom Moog) (2000-08-05)|
|From:||Tom Moog <firstname.lastname@example.org>|
|Date:||5 Aug 2000 21:37:42 -0400|
|References:||00-07-055 00-07-069 00-07-093 00-08-004|
As a long time (former) VMS user I can tell you that it is not VMS DCL
that it resembles, but the VMS Network Control Program (NCP) language
which I believe was created for RSX-11 (RIP) in the early 70s.
You cannot parse this language using traditional lexer/parsers without
hacks. It is not the grammar of the language which is the problem.
It is (a) the feedback from the parser to the lexer required to handle
matching of tokens by using the first unambiguous prefix and (b) the
need to supply descriptions of the allowed input at any point in the
parse, including different descriptions of the same token at different
points in the parse. For example:
set ip 126.96.36.199 subnet 255.255.255.240
The first a.b.c.d has a different help message than the second one.
Return to the
Search the comp.compilers archives again.