PLT Scheme v4.0.1

PLT Scheme version 4.0.1 is now available from http://plt-scheme.org/ This release fixes a number of bugs. Most noteworthy, the "Internal Error" message when requiring a PLaneT package has been fixed and user-defined teachpacks now work properly. Feedback Welcome.

PLT Scheme version 4.0 is now available from

http://plt-scheme.org/

This major new release offers many improvements over version 372, and we encourage everyone to upgrade.

• The PLT Scheme language now provides better syntax for modules, better support for optional and keyword arguments to functions, a more complete syntax for structure types, new syntax for list comprehensions and iterations, a more complete and consistent set of list operations, a more complete set of string operations, and streamlined hash-table operations.
• The documentation has been re-organized and re-written. New tutorials and overviews offer a clearer introduction to Scheme and PLT Scheme.
• New documentation tools help programmers create and install documentation for libraries and Planet packages. All installed documentation can be read though the user's web browser, and even searching within the browser works on local files.
  The language for writing documentation is an extension of Scheme, and document sources are linked to implementations through the module system. The module connection allows, for example, reliable automatic hyperlinking of identifiers mentioned in documentation to their specifications in other documentation.
• R6RS programs are supported in two ways: though the plt-r6rs executable and through the #!r6rs prefix. The latter allows an R6RS library or program to serve as a PLT Scheme module.
• Legacy R5RS support is improved, partly through a separate plt-r5rs executable.
• Pairs are immutable within the PLT Scheme language; mutable pairs (which are the same as R6RS and R5RS pairs) are provided as a separate datatype. For more information, see “Getting rid of set-car! and set-cdr!”
• ProfessorJ uses a new and improved parser, it evaluates programs faster, and it includes a Java-specific indenter.
• Testing frameworks for the HtDP and HtDC (ProfessorJ) teaching languages have been unified. Both support systematic unit testing in a comprehensive fashion. When programs lack tests, students are asked to add test cases. When all tests succeed, a simple message says so; otherwise, a pop-up window (dockable) displays URLs to the failed test cases and explains why the cases failed.
• Typed Scheme, a statically typed dialect of Scheme, is now included with PLT Scheme. While Typed Scheme is still in its early stages of development, it supports modular programming with types and full interaction with existing untyped code. Safe interactions between typed and untyped modules are enforced via contracts. Typed Scheme also features a novel type system designed to accommodate Scheme programming idioms. For more information, see the Typed Scheme page.

Feedback Welcome.

The Tour in Video

Putting behind its stodgy textual past, the DrScheme tour is now in video! It's rather preliminary, and can use lots of improvement, but now we're really taking it to the kids.

PLT Scheme version 4.0 is Coming Soon

PLT Scheme is now 13 years old. The initial version was little more than glue code between a few open-source libraries, which seemed to offer the quickest solution to our modest goals. Modest success leads to bigger goals, however, and then continued success leads to ever more ambitious goals. Before you know it, a mass of users, co-developers, libraries, and documentation rely on design decisions that were made for a much smaller project years before.

Naturally, many of those early design decisions turn out to be a poor fit for the project’s eventual role. Starting from scratch isn’t usually practical, so you gradually adjust the infrastructure to meet new needs. That was precisely the story for the version 300 series of releases for PLT Scheme. The biggest gap between our original and current goals was in run-time performance, so we replaced bytecode interpretation with a just-in-time native-code compiler, and we replaced a memory manager based on “conservative” estimates of pointer usage with one that uses precise information.

Performance improves a bit more with version 4.0, but mostly we’ve moved on to a bigger mismatch between the original and current goals: the way that PLT Scheme presents itself to users. PLT Scheme was originally conceived as R5RS Scheme with some extensions to make it practical, and with useful tools (notably an IDE) and libraries (notably a GUI library) built on that core. Our documentation and web pages reflected that architecture – which now seems completely upside-down.

Version 4.0 is a fresh start in the way that we present PLT Scheme. It’s a new language. PLT Scheme is a dialect of Scheme, certainly, but it’s not merely a superset of R5RS, R6RS, or other standards, and those standards are not really the best place to start understanding PLT Scheme. At the same time, the unique extensibility features of the PLT Scheme language and tools allow them to support other languages easily, including R5RS (though a new plt-r5rs executable), R6RS, and more.

Improvements to the PLT Scheme language include better syntax for modules, better support for optional and keyword function arguments, more expressive syntax for structure types, streamlined hash-table operations, new syntax for list comprehensions and iterations, a more complete and consistent set of list and string operations, and reduced dependence on mutable pairs. To current users of PLT Scheme, these changes will seem like the big ones behind version 4.0, but they’re small compared to the overall re-organization and the accompanying documentation effort.

We wrote hundreds of pages of new documentation, including much more tutorial and overview information. We ported hundreds of pages of existing documents to new a system that produces cleaner, better organized, more consistent output. We will replace the old tangle of web pages (that try to explain a confusing federation of tools) with a simple page about “PLT Scheme.” We have even streamlined the command-line flags for the main virtual machine.

The development of PLT Scheme version 4.0 took about one year of hard work. In retrospect, that doesn’t sound too bad, considering the scale of the existing code base, the number of things that we improved, and the total size of the documentation (about 2000 pages in PDF form). Still, you can imagine how happy we are to arrive at a stable release, and we hope that the improvements in PLT Scheme version 4.0 work as well for everyone else as they do for us.

For a preview, see http://pre.plt-scheme.org/. The final version 4.0 release is just days away.