Python in the browser, precompiled for speed: http://www.transcrypt.org
- Multiple inheritance, optional operator overloading, metaclasses, properties, class and function decorators, hierarchical modules etc.
- Pure Python 3.5 syntax, using Python’s native parser
- Debug directly from Python sourcecode, through integrated sourcemaps
- Integrated static typechecking and minification at the tip of a command line switch
- Also runs on top of node.js
- Extensive documentation with many code examples
- Apache 2.0 license
- Pip-install and go!
Latest stable release: Paris
To obtain the latest stable release including all updates, install it from PyPi as described in the documentation.
Thanks to everyone who contributed!
Main differences with CPython
- A few methods of builtin types are currently left out, especially when they (almost) duplicate functionality of other methods. Also method decorators (as opposed to function decorators and class decorators) are not supported, with the exception of @classmethod. This results from a deliberate choice to keep Transcrypt lean and fast.
Information for contributors
How to contribute
Transcrypt started out as a personal repo, owned by Jacques de Hooge. As the project caught on and the number of people contributing issues, ideas and code grew, the repo was transferred to the QQuick organisation, to be able to form a developer team on GitHub.
There was also a clear message in this: Transcrypt isn’t owned by anyone in particular. It is the collective property of everyone using it or contributing to it. At the same time the need was felt to keep a very firm grip on code quality, especially of the core.
Everything under ../transcrypt/modules/org/transcrypt plus the file ../transcrypt/__main__.py is considered to be part of Transcrypt’s core. A major design goal is to keep the core small and fast. This means that some CPython facilities were deliberately left out or simplified. Core development is still mainly done by Jacques, but with the input of many great ideas submitted as issues. If you want to improve something in the core, this is best initiated by first opening an issue for it. Opening a pull request directly can lead to disappointment, although all effort is made to take good ideas seriously.
All other parts of Transcrypt are referred to as periphery. While a good quality pull request for the periphery stands a reasonable chance of being accepted, still it is wise to start an issue beforehand, allowing coordination and preventing waste of effort.
While being open and respectful to any good ideas, the final say as to what gets in and what doesn’t, is with Jacques. So this is a dictatorial rather than a democratic project. Being a sailer himself, Jacques values the notion of having one captain on a ship. The captain doesn’t own the ship, but he serves the passengers by consulting with the crew and plotting one stable course.
Another possibility to contribute libraries to Transcrypt is by submitting them as separate packages to PyPi. In that case be sure to add the keyword Transcrypt to allow people to find your package. Making your package pip-installable will also help it to catch on. Contributing packages via PyPi of course means total freedom for the developer.
Development build status
The icon above shows the outcome of the continous integration test that is done on Linux after each commit. The test consists of running a set of testlets, systematically covering all facilities of Transcrypt. Each testlet performs an automated back to back test against CPython.
The full set of testlets is described in the documentation and comes with the distribution. Since the branching model has been deliberately kept simple, continuous integration may be transiently broken. In that case you can use the latest passing version, that you’ll find by clicking on the icon above and then on ‘Build History’.
Each release, on the other hand, is preceded by at least the following tests:
- The automated back to back test described above, not only on Linux but also on Windows and, in case of relevant issues, on OsX.
- Automated compilation of the manual tests, human exercising of the resulting applications and a visual check of the results.
- Automated compilation of the demo’s, human exercising of the resulting applications and a visual check of the results.
- A documentation build, followed by a visual sample check.
What’s new in the latest commits
- Path search order fixed
- Regular expressions autotest workaround added for Python 3.6 enums
- Async/await added (EXPERIMENTAL) + manual test
- Made Python 3.6 the default in several places
- Pragma and command line switch added to augment module search path
- Better example of use of __getattr__ and __setattr__ added to autotest suite
- Method dict.values () added
- Demo added for three.js with encapsulated constructors
- Doc link repaired
- Fix for #317: Wrong exception type for aList [‘aStringLiteral’]
- Enhancement for #316: filter doesn’t support None for func
- Enhancement for #314: float (‘ ‘) returns 0
- Enhancement for #310: hasattr () raises ‘Uncaught TypeError’
- Enhancement for # 306: dict.popitem () added + autotest
- Integrated with newest API of mypy
- Fix for issue #304: Invalid JS when using global –opov flag
- Fix for issue #301: zip broken for strings
- Comment-like pragma’s (issue #295) added: # __pragma__ (<parameters>)
- Fix for issue #284 (+= problem) and autotest case
- Bundled version 0.4.4 of mypy static typechecker replaced by dependency on mypy (currently 0.4.7)
- Added in-place overloads for @=, **=, %=, *=, /=, +=, -=, <<=, >>=, |=, ^=
- Operator **= added, a **= b converted to a = a ** b
- Operator @= added, M3 @= M2 converted to M3 = M3 @ M2, same as for other augmented assignment ops
- Compilation error report now contains detailed import sequence
- Enhancement for issue #281: tuple keys allowed for dictionaries
- Enhancement for issue #26: super () added for unique path to single ancestor method
- Fix for issue #279: TypeError exists both in Python and JavaScrip, needs alias
- Fix for issue #277: Alias needed for new
- Fix for issue #274: Cannot delete unqualified property in strict mode
- Fix for issue #268: Module import trace missing in error messages
- Autotest for regular expression module made part of shipment test and CI test. Some parts commented out, marked with @JdeH
- Documentation updated, also on-line
- Fix for issue #256: Parenthesis aren’t translated (before dot, e.g. in return)
- Option -b (or –blind) added to shipment test. Running it with -c (or –clean) -b (or –blind) will not show anything, just clean repo by removing all generated files. Meant for usage before a commit.
- [Release Paris (PyPi v3.6.4, GitHub #14): Support for Python 3.6, div. fixes and enhancements]
- Function globals () is now available to dynamically get and set module attribs, issue #251
- Demo for cyclejs was added, also to the shipment test
- Enhancement for issue #247: One web page can now hold multiple Transcrypt apps
- Regular expression module finished! (issue #98)
- Fixes for issues #254 and #252, both having to do with sourcemap being one line off in else clause
- Emulation of sync console I/O for educational purposes, text only
- Autotest output now in tabular form for easy comparison, incl. source line nrs
- Fix for issue #178: List Comprehensions / Operator Precedence Bug
- Div. enhancements and fixes for CI tests
- Enhancement for issue #139: ‘yield from’ now supported.
- Enhancement for issue #89 and #149: __getattr__ and __setattr__ are now supported, requiring the -e 6 switch. Testlet ‘proxies’ added.
- New aliases added to prevent name clashes. The orignal name can always be reached by prepending js_. So e.g. if you need ‘clear’ in JS, use ‘js_clear’ in Python. A complete list of aliases is in the docs. Any alias can be undefined to maintain backward compatibility, e.g __pragma__ (‘noalias’, ‘clear’).
- Enhancement for issue #169: Add support for float(‘inf’) and float(‘-inf’)
Known bugs in latest commits
Other packages you might like
- Multi-module Python source code obfuscator: https://pypi.python.org/pypi/Opy
- PLC simulator with Arduino code generation: https://pypi.python.org/pypi/SimPyLC
- A lightweight Python course taking beginners seriously (under construction): https://pypi.python.org/pypi/LightOn
- Event driven evaluation nodes: https://pypi.python.org/pypi/Eden
Copyright 2014, 2015, 2016 Jacques de Hooge, GEATEC engineering, www.geatec.com
Licensed under the Apache License, Version 2.0 (the “License”); you may not use this file except in compliance with the License. You may obtain a copy of the License at
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an “AS IS” BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.