amber/CHANGELOG

 2013 - Release 0.12.0
===============================

After 3 months of work we are pleased to announce the 0.12.0 release of Amber.

Besides the usual bug fixes a lot of new features have emerged.

The biggest change is the switch to RequireJS to specify Amber package dependencies
and for loading the packages as AMD modules.
Amber is now additionally available as Bower [2] component.
Bower is also used to manage required JavaScript libraries which don't have to be kept
around in the repository anymore.

The repository layout was restructured to provide a cleaner separation of different Amber parts:
* Smalltalk code is located in 'st'
* Compiled Amber packages are located in 'js'
* Supporting JavaScript code is located in 'support'

Together with the RequireJS changes the specifying their dependencies the Brickz [3]
reconfigurable micro composition system was introduced.
This is used for

On the Smalltalk side support has been added for writing exponential numbers of the form 2e5.

Helios (the new IDE) is progressing nicely and has seen a lot of improvements.
One of the great parts is the new stepping debugger which is also making progress.
To try Helios, open the helios.html page or
evaluate in any other amber page `require('amber/helpers).popupHelios()`.

The last enhancements target the commandline compiler which can be used as
`amberc` (an executable script) or as a Grunt task.
The following features have been added:
 * generation of shebang line (#!/usr/bin/env node)
 * specification of target namespace (`-n` flag; `amd_namespace` option in Grunt)
 * specification of additional library directories (`-L` flag; `library_dirs` option in Grunt)
The following features have been removed:
 * creation of *.deploy.js files (`-d` flag; `deploy` option in Grunt)
 * optimization passes via Closure compiler (`-o`/`-O` flags; `closure_jar` option in Grunt)
   The same behavior can be achieved by using specific Grunt tasks

Additionally, the Grunt task now handles the -v/--verbose flag which triggers the same behavior
as the `verbose` option which can be specified in the Gruntfile.


Some numbers about this release (starting from 0.11.0):

* 660 commits
* 10 committers
* 66 unit tests added
* 152 issues were closed
* 379 unit tests in total


Commits: https://github.com/amber-smalltalk/amber/compare/0.11.0...0.12.0
Issues: https://github.com/amber-smalltalk/amber/issues?direction=desc&milestone=8&page=1&sort=updated&state=closed

For the most important API related changes see the file API-CHANGES.txt.


* Installing Amber from NPM

To install Amber from NPM, run

    npm install amber

* Installing Amber from Bower

To install Amber from Bower, run

    bower install amber

* Migration from Amber 0.11.0

First, the loading of JavaScript files must be adapted. The custom loader has been replaced with
requirejs for loading files in the browser. New loader code is thouroughly
explained in [4].

After updating the loader part, `.st` files need to be recompiled
into new AMD `.js` files. During loader changes, a namespace was choosen and will be needed for recompilation.

Go to your directory with `.st` files and issue this from the shell:

```sh
<path-to-amber>/bin/amberc -l SUnit,Canvas -n <chosen-namespace> -D <path-for-compiled-js-files> *.st
```

In windows, use `\` as path separator, the rest is identical.

The `-l SUnit,Canvas` is just a general guess, if your code depends on more (or less) non-kernel packages from amber, list them here accordingly.

This migrate scenario only covers simple deployments with your project's code and amber.
If your project is more compilicated, using libraries and packages from multiple sources,
it is hard to give general advices to migrate - you must do it library by library,
giving each location its own namespace, and `-L` option of `amberc`
comes handy when integrating; ask on the mailing list if problems arise.

[1] http://requirejs.org/
[2] http://bower.io/
[3] https://github.com/amber-smalltalk/brikz
[4] https://github.com/amber-smalltalk/amber/wiki/How-to-load-amber


09th July 2013 - Release 0.11.0
===============================

Three months have passed and we are happy to announce the release
of Amber 0.11.0!

Some numbers about this release:

* 494 commits
* 13 committers
* increasing the number of core committers to 25
* 50 unit tests added to the kernel
* 313 unit tests in total

Since the last release 60 issues were closed, bringing us to 499
issues closed.

This release includes a lot of bug fixes, improvements to the
CLI, as well as a preview of the next IDE, named Helios. Amber
now also uses a CI server [1].

To try the Helios, the new IDE, open the helios.html page, or
evaluate in any other amber page `amber.loadHelios()`.

The compiler also received some improvements, especially
regarding message send optimisations and super sends.

Here's the list of commits and closed issues:
https://github.com/amber-smalltalk/amber/compare/0.10.0...0.11.0
https://github.com/amber-smalltalk/amber/issues?direction=desc&milestone=6&page=1&sort=updated&state=closed

There is a lot more to come with Helios, a stepping debugger
based on the AST is in progress and currently in the 'debugger'
branch on GitHub.

* Installing Amber from NPM

To install amber from NPM, run

    npm install amber


* Migration from Amber 0.10.0

Amber compiled code should be compatible with Amber 0.10.0, but
recompiling is needed in order to get contexts working:

    Compiler new recompile: (Package named: 'MyPackage') classes.
    (Package named: 'MyPackage') commit

For API related changes see the file API-CHANGES.txt.


13th March 2013 - Release 0.10.0
================================

Here's a summary of change since the 0.9.1 release:

- Travis CI jobs
- Almost 300 issues closed
- 150 new unit tests written
- All classes in the Kernel are documented

- New and much improved compiler toolchain, providing:
  - Semantic analysis
  - AST Node annotation
  - Intermediate representation (easier to optimize)
  - Better inlining
  - an AST interpreter
  - support for blocks contexts

- New build system based on grunt.js, replacing the old makefiles
- New bin/amberc compiler written in Amber/Node.js
- SUnit improvements and cleanup, with support for async assertions
- Improved ClassBuilder with better class migration support
- Improved loader for third party packages
- New: Announcements
- Classes can be augmented by light-weight methods created from blocks
- Snippets of HTML can be marked to become virtual tags in HTMLCanvas
- Amber server supports Basic HTTP authentication (not recommended for production environments/unencrypted connections)
- New IDE on it's way for Amber 1.0



16th January 2012 - Release 0.9.1
=================================

Here's a summary of changes since the 0.9 release:

- 80 new unit tests written
- 52 issues fixed
- All classes in Kernel-Objects, Kernel-Classes and Kernel-Methods has been documented
- New documentation framework (see http://amber-lang.net/documentation.html)
- Better class organisations, "Kernel" package split into several packages
- First class packages have replaced class categories
- Internet Explorer 7+ compatibility
- New Announcement framework ported from Pharo
- New console-based REPL written in Amber using node.js
- Symbol class implemented together with object identity and #==
- New OrderedCollection and Set implementation
- Dictionary can now have any kind of object as keys. String-key dictionary has been renamed HashedCollection
- New TwitterWall example
- Improved HTML Canvas, now compatible with IE7
- Improved JSObjectProxy for seamless JavaScript objects access from Amber
- No more jQuery binding. Amber is fully capable of sending messages to JavaScript objects

13th September 2011 - Release 0.9
=================================
Amber has been evolving furiously since the presentation at ESUG 2011 in Edinburgh less than 3 weeks ago.

This is a summary:

Language, compiler and runtime

- New 100x faster parser built using PEGjs instead of the old parser built using PetitParser.
- New much faster ChunkParser implementation in handwritten Amber instead of using PetitParser.
- Improved parsing error report with quoted source code plus marker pinpointing parsing error.
- Removed PetitParser since it is no longer needed by Amber itself.
- Added compiler optimizations in the form of speculative inlining of specific messages and control structures.
- Added support for dynamic Arrays, just like in Squeak/Pharo.
- Added support for similar dynamic Dictionaries, not seen in other Smalltalks.
- Added & and | as allowed binary selectors and implemented them in Boolean.
- Added a Set implementation.
- Added first basic support for real Packages with dependency information.

...and various extensions, enhancements and bug fixes to the library classes.


Development environment

- A working Debugger with integrated inspector, proceed etc.
- A new structure with
- A working amberc command line compiler including a Makefile for recompiling the whole Amber.
- Enabled TestRunner in the IDE for running unit tests based on SUnit.
- Added "File in" button in Workspace to easily paste and filein source code in chunk format in the IDE.


Example code and ports

- Ported ProfStef interactive tutorial, available on Amber homepage but also in examples directory.
- Included the ESUG presentation as an example also in the examples directory.
- Several new examples running on Node.js and webOS included, all with Makefiles.


Various other things

- Issue tracker on github now used as primary source, closed a bunch of reported issues.
- Wiki pages on github with information on how to port code from other Smalltalks, lists of articles, tutorials, roadmap and more.