EDITOR-HINTS.NAMED-READTABLES

EDITOR-HINTS.NAMED-READTABLES - 0.9

by Tobias C Rittweiler

Repository:

darcs get http://common-lisp.net/project/editor-hints/darcs/named-readtables/

Download:

editor-hints.named-readtables-0.9.tar.gz

What are Named-Readtables?
Notes on the API
Important API idiosyncrasies
Preregistered Readtables
Examples
Acknowledgements
Dictionary

What are Named-Readtables?

Named-Readtables is a library that provides a namespace for readtables akin to the
already-existing namespace of packages. In particular:

you can associate readtables with names, and retrieve readtables by names;
you can associate source files with readtable names, and be sure that the right readtable is
active when compiling/loading the file;
similiarly, your development environment now has a chance to automatically determine what
readtable should be active while processing source forms on interactive commands. (E.g. think
of `C-c C-c' in Slime [yet to be done])

Additionally, it also attempts to become a facility for using readtables in a modular way. In
particular:

it provides a macro to specify the content of a readtable at a glance;
it makes it possible to use multiple inheritance between readtables.

The API heavily imitates the API of packages. This has the nice property that any experienced
Common Lisper will take it up without effort.

DEFREADTABLE - DEFPACKAGE

IN-READTABLE - IN-PACKAGE

MERGE-READTABLES-INTO - USE-PACKAGE

MAKE-READTABLE - MAKE-PACKAGE

UNREGISTER-READTABLE - DELETE-PACKAGE

RENAME-READTABLE - RENAME-PACKAGE

FIND-READTABLE - FIND-PACKAGE

READTABLE-NAME - PACKAGE-NAME

LIST-ALL-NAMED-READTABLES - LIST-ALL-PACKAGES

Important API idiosyncrasies

There are three major differences between the API of Named-Readtables, and the API of packages.

1. Readtable names are symbols not strings.

Time has shown that the fact that packages are named by strings causes severe headache because of
the potential of package names colliding with each other.

Hence, readtables are named by symbols lest to make the situation worse than it already is.
Consequently, readtables named CL-ORACLE:SQL-SYNTAX and CL-MYSQL:SQL-SYNTAX can happily coexist
next to each other. Or, taken to an extreme, SCHEME:SYNTAX and ELISP:SYNTAX.

If, for example to duly signify the importance of your cool readtable hack, you really think it
deserves a global name, you can always resort to keywords.

2. The inheritance is resolved statically, not dynamically.

A package that uses another package will have access to all the other package's exported
symbols, even to those that will be added after its definition. I.e. the inheritance is resolved at
run-time, that is dynamically.

Unfortunately, we cannot do the same for readtables in a portable manner.

Therefore, we do not talk about "using" another readtable but about "merging"
the other readtable's definition into the readtable we are going to define. I.e. the
inheritance is resolved once at definition time, that is statically.

(Such merging can more or less be implemented portably albeit at a certain cost. Most of the time,
this cost manifests itself at the time a readtable is defined, i.e. once at compile-time, so it may
not bother you. Nonetheless, we provide extra support for Sbcl, ClozureCL, and AllegroCL at the
moment. Patches for your implementation of choice are welcome, of course.)

3. DEFREADTABLE does not have compile-time effects.

If you define a package via DEFPACKAGE, you can make that package the currently active package for
the subsequent compilation of the same file via IN-PACKAGE. The same is, however, not true for
DEFREADTABLE and IN-READTABLE for the following reason:

It's unlikely that the need for special reader-macros arises for a problem which can be
solved in just one file. Most often, you're going to define the reader macro functions, and
set up the corresponding readtable in an extra file.

If DEFREADTABLE had compile-time effects, you'd have to wrap each definition of a
reader-macro function in an EVAL-WHEN to make its definition available at compile-time. Because
that's simply not the common case, DEFREADTABLE does not have a compile-time effect.

If you want to use a readtable within the same file as its definition, wrap the DEFREADTABLE and
the reader-macro function definitions in an explicit EVAL-WHEN.

Preregistered Readtables

- NIL, :STANDARD, and :COMMON-LISP designate the standard readtable.

- :MODERN designates a case-preserving standard-readtable.

- :CURRENT designates the current readtable.

Examples

     (defreadtable elisp:syntax
        (:merge :standard)
        (:macro-char #\? #'elisp::read-character-literal t)
        (:macro-char #\[ #'elisp::read-vector-literal t)
        ...
        (:case :preserve))
    
     (defreadtable scheme:syntax
        (:merge :standard)
        (:macro-char #\[ #'(lambda (stream char)
                              (read-delimited-list #\] stream)))
        (:macro-char #\# :dispatch)
        (:dispatch-macro-char #\# #\t #'scheme::read-#t)
        (:dispatch-macro-char #\# #\f #'scheme::read-#f)
        ...
        (:case :preserve))
    
     (in-readtable elisp:syntax)
    
     ...
    
     (in-readtable scheme:syntax)
    
     ...

Acknowledgements

Thanks to Robert Goldman for making me want to write this library.

Thanks to Stephen Compall, Ariel Badichi, David Lichteblau, Bart Botta, David Crawford, and Pascal
Costanza for being early adopters, providing comments and bugfixes.

Dictionary

[Function]
copy-named-readtable named-readtable => result

Argument and Values:

named-readtable: (OR READTABLE SYMBOL)

result: READTABLE

Description:

Like COPY-READTABLE but takes a NAMED-READTABLE-DESIGNATOR as argument.

[Macro]
defreadtable name &body options => result

Description:

Define a new named readtable, whose name is given by the symbol name. Or, if a readtable is
already registered under that name, redefine that one.

The readtable can be populated using the following options:

(:MERGE readtable-designators+)

Merge the readtables designated into the new readtable being defined as per MERGE-READTABLES-INTO.

If no :MERGE clause is given, an empty readtable is used. See MAKE-READTABLE.

(:FUZE readtable-designators+)

Like :MERGE except:

Error conditions of type READER-MACRO-CONFLICT that are signaled during the merge operation will
be silently continued. It follows that reader macros in earlier entries will be overwritten by
later ones.

(:DISPATCH-MACRO-CHAR macro-char sub-char function)

Define a new sub character sub-char for the dispatching macro character macro-char,
per SET-DISPATCH-MACRO-CHARACTER. You probably have to define macro-char as a dispatching
macro character by the following option first.

(:MACRO-CHAR macro-char function [non-terminating-p])

Define a new macro character in the readtable, per SET-MACRO-CHARACTER. If function is the
keyword :DISPATCH, macro-char is made a dispatching macro character, per
MAKE-DISPATCH-MACRO-CHARACTER.

(:SYNTAX-FROM from-readtable-designator from-char to-char)

Set the character syntax of to-char in the readtable being defined to the same syntax as
from-char as per SET-SYNTAX-FROM-CHAR.

(:CASE case-mode)

Defines the case sensitivity mode of the resulting readtable.

Any number of option clauses may appear. The options are grouped by their type, but in each group
the order the options appeared textually is preserved. The following groups exist and are executed
in the following order: :MERGE and :FUZE (one group), :CASE, :MACRO-CHAR and :DISPATCH-MACRO-CHAR
(one group), finally :SYNTAX-FROM.

Notes:

The readtable is defined at load-time. If you want to have it available at compilation time -- say
to use its reader-macros in the same file as its definition -- you have to wrap the DEFREADTABLE
form in an explicit EVAL-WHEN.

On redefinition, the target readtable is made empty first before it's refilled according to
the clauses.

NIL, :STANDARD, :COMMON-LISP, :MODERN, and :CURRENT are preregistered readtable names.

[Function]
ensure-readtable name &optional default => result

Argument and Values:

name: (OR READTABLE SYMBOL)

default: (OR READTABLE SYMBOL)

result: READTABLE

Description:

Looks up the readtable specified by name and returns it if it's found. If it is not
found, it registers the readtable designated by default under the name represented by
name; or if no default argument is given, it signals an error of type
READTABLE-DOES-NOT-EXIST instead.

[Function]
find-readtable name => result

Argument and Values:

name: (OR READTABLE SYMBOL)

result: (OR READTABLE NULL)

Description:

Looks for the readtable specified by name and returns it if it is found. Returns NIL
otherwise.

[Macro]
in-readtable name => result

Description:

Set *READTABLE* to the readtable referred to by the symbol name.

[Function]
list-all-named-readtables => result

Argument and Values:

result: LIST

Description:

Returns a list of all registered readtables. The returned list is guaranteed to be fresh, but may
contain duplicates.

[Function]
make-readtable &optional name &key merge => result

Argument and Values:

name: (OR READTABLE SYMBOL)

merge: LIST

result: READTABLE

Description:

Creates and returns a new readtable under the specified name.

merge takes a list of NAMED-READTABLE-DESIGNATORS and specifies the readtables the new
readtable is created from. (See the :MERGE clause of DEFREADTABLE for details.)

If merge is NIL, an empty readtable is used instead.

If name is not given, an anonymous empty readtable is returned.

Notes:

An empty readtable is a readtable where each character's syntax is the same as in the
standard readtable except that each macro character has been made a constituent. Basically:
whitespace stays whitespace, everything else is constituent.

[Function]
merge-readtables-into result-readtable &rest named-readtables => result

Argument and Values:

result-readtable: (OR READTABLE SYMBOL)

named-readtables: (OR READTABLE SYMBOL)

result: READTABLE

Description:

Copy the contents of each readtable in named-readtables into result-table.

If a macro character appears in more than one of the readtables, i.e. if a conflict is discovered
during the merge, an error of type READER-MACRO-CONFLICT is signaled.

[Type]
named-readtable-designator

Description:

Either a symbol or a readtable itself.

[Condition type]
reader-macro-conflict

Description:

Continuable.

This condition is signaled during the merge process if a) a reader macro (be it a macro character
or the sub character of a dispatch macro character) is both present in the source as well as the
target readtable, and b) if and only if the two respective reader macro functions differ.

[Condition type]
readtable-does-already-exist

Description:

Continuable.

[Condition type]
readtable-does-not-exist

[Function]
readtable-name named-readtable => result

Argument and Values:

named-readtable: (OR READTABLE SYMBOL)

result: SYMBOL

Description:

Returns the name of the readtable designated by named-readtable, or NIL.

[Function]
register-readtable name readtable => result

Argument and Values:

name: SYMBOL

readtable: READTABLE

result: READTABLE

Description:

Associate readtable with name. Returns the readtable.

[Function]
rename-readtable old-name new-name => result

Argument and Values:

old-name: (OR READTABLE SYMBOL)

new-name: SYMBOL

result: READTABLE

Description:

Replaces the associated name of the readtable designated by old-name with new-name.
If a readtable is already registered under new-name, an error of type
READTABLE-DOES-ALREADY-EXIST is signaled.

[Function]
unregister-readtable named-readtable => result

Argument and Values:

named-readtable: (OR READTABLE SYMBOL)

result: (MEMBER T NIL)

Description:

Remove the association of named-readtable. Returns T if successfull, NIL otherwise.

This documentation was generated on 2009-11-5 from a Lisp image using some home-brewn, duct-taped,
evolutionary hacked extension of Edi Weitz' DOCUMENTATION-TEMPLATE.