DataMuseum.dk

Presents historical artifacts from the history of:

DKUUG/EUUG Conference tapes

This is an automatic "excavation" of a thematic subset of
artifacts from Datamuseum.dk's BitArchive.

See our Wiki for more about DKUUG/EUUG Conference tapes

Excavated with: AutoArchaeologist - Free & Open Source Software.


top - metrics - download
Index: D T

⟦bd451d89b⟧ TextFile

    Length: 4429 (0x114d)
    Types: TextFile
    Names: »DESCRIPTION«

Derivation

└─⟦52210d11f⟧ Bits:30007239 EUUGD2: TeX 3 1992-12
    └─⟦c319c2751⟧ »unix3.0/TeX3.0.tar.Z« 
        └─⟦036c765ac⟧ 
            └─⟦this⟧ »TeX3.0/Spiderweb/doc/DESCRIPTION« 
└─⟦52210d11f⟧ Bits:30007239 EUUGD2: TeX 3 1992-12
    └─⟦63303ae94⟧ »unix3.14/TeX3.14.tar.Z« 
        └─⟦c58930e5c⟧ 
            └─⟦this⟧ »TeX3.14/Spiderweb/doc/DESCRIPTION« 

TextFile

Until recently, the only WEB implementations I have been aware of have
been Knuth's original Pascal implementation, and some ports to C and
Modula-2.  I have developed a new tool, Spider, which reads a
description of a programming language and writes a WEB system that can
be used to write programs in that language.  My colleagues and I have
used Spider to build WEB systems for Ada, C, AWK, and other languages.
I am now making a beta test version of Spider available for anonymous
ftp.  A description follows.


Spider reads a description of a programming language, and writes
source code for a Weave and Tangle which support that language.  The
language-dependent parts are derived from the description given
Spider, and the language-independent parts are taken from ``master
copies'' of Weave and Tangle.  The master copies are derived from
Silvio Levy's CWEB, and the Tangle and Weave that are constructed are
implemented in C.  Spider is implemented as an Awk program.

An exhaustive list of Spidery Web's features would interest only Web
experts, but I do want to mention some features that I hope will
encourage people to use Spidery {\tt WEB}.

	Tangle and Weave can read from multiple files (this feature is 
	present in Levy's CWEB), and Tangle can write to multiple files.
	Included files will be searched for on a path if not found in the
	current directory.  These features make Spidery WEB more
	usable on systems that have make.

	TANGLE can expand macros with multiple parameters.

	TANGLE writes #line directives, so you can debug at the	WEB
	source level if your compiler respects the C~conventions for #line.


Many features of WEB seem to exist only to compensate for deficiencies
in PASCAL, and most of those were dropped in CWEB.  I have changed
much of CWEB in order to avoid being bound too much by C conventions.
As a result, there are dozens of minor differences between Spidery WEB
and original WEB.  To give just one example, Spidery WEB supports
octal and hexadecimal constants using WEB-style notation, not the C
notation used in CWEB.

Spider can generate WEB systems for a variety of languages.  The
author has written Spider description files for C, AWK, Ada, SSL (a
language that describes attribute grammars to the Cornell Synthesizer
Generator), the Larch Shared Language (a language for describing
equational theories), and Dijkstra's language of guarded commands.
Debugging the grammar that WEAVE uses to prettyprint the language is
the most time-consuming part of creating a WEB system for a new target
language, and Spider makes it trivial to change that grammar.  To make
a Spider description file for an Algol-like language that uses infix
expression notation, an experienced systems programmer should be able
to adapt an existing Spider description file very quickly.

Spider's major limitations are lexical.  All Spidery WEBs assume that
spaces and tabs in the input are not significant, except as
separators; this makes it impossible to construct Spidery WEBs for
languages like Fortran and Miranda, where the position of text on a
line is significant.  (You would have to modify the ``master copies''
of Tangle and Weave to change the lexical analyzer.)  The lexical
structures of identifiers, string literals, and numeric literals are
fixed.

Spider's major value is that is makes it possible to create a new WEB
quickly, and to tinker with it easily.  My colleagues and I have used
Spidery WEB to write programs in Ada, C, and SSL, and has been pleased
with the result.  We have written in WEB an application of 24 thousand
lines, and we are very pleased at how easy it has been to review and
maintain this code.  I hope that the availability of Spidery WEB will
encourage other groups to try literate programming, and that they,
too, will be pleased with the results.


Spidery Web is available by anonymous ftp from princeton.edu, in file
	~ftp/pub/spiderweb.tar.Z 
in compressed tar format.  I have tested this system only under Unix.
While there is no reason to believe it won't work on other systems, it
does require a C compiler and an Awk filter.  I would like to here
about experience with other systems, especially MS-DOS.  There is a
good public-domain awk (AWK210.ARC) available by anonymous ftp from
grape.ecs.clarkson.edu, but my PC doesn't have enough processing power
to make it worth the attempt.

Send comments or bug reports to nr@princeton.edu.

Norman Ramsey