Permalink
Browse files

Update the README and History files.

  • Loading branch information...
1 parent fb515a7 commit 5adb3608bf3415bd265df51df80dfbdeedd7ba24 @jcoglan committed Aug 23, 2009
Showing with 29 additions and 4 deletions.
  1. +7 −0 History.txt
  2. +22 −4 README.rdoc
View
@@ -1,3 +1,10 @@
+=== Version 0.3.1 (2009-08-23)
+
+* Added the (rationalize) procedure
+* Allowed (gcd) and (lcm) to take more than two arguments
+* Fixed string output so the REPL prints all data types correctly
+
+
=== Version 0.3.0 (2009-07-31)
We've got all your data types now
View
@@ -38,7 +38,7 @@ languages.
Heist nominally targets R5RS (http://www.schemers.org/Documents/Standards/R5RS/HTML).
It has good support for many Scheme runtime features such as tail call
optimisation, macros and first-class continuations, as well as reasonably
-complete numeric and list libraries.
+complete libraries for all the Scheme data types.
Currently implemented R5RS features include:
@@ -228,18 +228,34 @@ explicitly documented in this file can be assumed to work according to
the Scheme standard, or according to the information presented here,
and can be assumed to be reasonably stable.
+=== Ruby-based syntax
+
I have not documented how to write your own syntax using Ruby because
it requires far too much knowledge of Heist's plumbing at present (I
suspect this may be unavoidable). Besides, we have macros so if you
want new syntax we've got you covered. In fact, writing syntax using
macros makes sure that new syntactic forms support continuations
correctly, and Heist itself eschews Ruby-based syntax where possible.
+=== Valid symbols
+
Heist is extremely liberal as regards symbols. Any sequence of
characters that does not contain any spaces or parentheses and that
-is not a boolean literal, a number or a string is considered a valid
-symbol. This means that syntax that should be used for chars and
-vectors will currently parse as symbols.
+is not a boolean literal, a number, a character literal or a string
+is considered a valid symbol.
+
+=== Vectors and quoting
+
+Vectors do not need to be quoted, for example the expression <tt>#(1 2 3)</tt>
+evaluates to itself. If a vector is quoted, it is made immutable and
+a quoted vector expression returns the same in-memory object every time
+it is evaluated. In contrast, an unquoted vector returns a new vector
+object every time it is evaluated. Unquoted vectors are not frozen
+as this makes macro transformations hard to implement, and because we
+cannot let the <tt>vector-set!</tt> procedure modify the parse tree
+each unquoted vector is copied on evaluation.
+
+=== Macros
Macros are slightly more liberal than R5RS, in that this is a valid
macro:
@@ -276,6 +292,8 @@ conditionally, certainly this will not work with macros as the
inlining process will overwrite conditional code using the first
expansion generated.
+=== Laziness
+
Lazy evaluation mode is mainly useful for doing lambda calculus
without being concerned about the difference between normal and
applicative order, which for example affects the expression for

0 comments on commit 5adb360

Please sign in to comment.