Q1 2015 - 3.2 (Language versioning, generator improvements, cross-model generation)
Q3 2015 - 3.5 (Web-based projectional editors)
Q2 2016 - 4.0 (Improved projectional editor with partial parsing support. New, simplified languages for DSL design)
The main goal is to allow for seamless language versioning and migrations between language versions.
MPS will support the notion of a language version. Migrations scripts will automatically migrate client's code to the most recent version of the used languages.
The MPS generator will be further optimised for performance. The generation plan creation as well as generation priorities will be reconsidered.
- Optional stripping of language definition implementation from generated language artifacts
- Fluent parentheses editing in BaseLanguage expressions
- Allow JDK8 as a runtime platform for MPS
- In-process execution mode for tests
The MPS projectional editor will be enabled to run in the web browser and thus allow to distribute MPS languages through the web.
The language-definition languages will get a face-lift to reflect the feedback we've received from the users and the progress we’ve made over time in terms of usability and intuitiveness.
- new query language that will not depend on Base Language
- new generator language
- new editor definition language (a higher-level than now, based on the new query language)
MPS already introduced table notations in the editor. Now we will provide a higher-level API for specifying tabular notations, and will provide editing usability improvements for tables.
It is currently not possible to project nodes into the editor when not owned by the current context node. This will become possible in 4.0: A concept will be able to declare "transient nodes", i.e. children that are computed based on DSL definition, which will be able to be projected as part of an editor definition.
- editor extension points to allow easier editor reuse in derived languages
- Introduce view-specific (or editor-specific) data into MPS so concept editors can save view-specific information
Sometimes, plain text, which feels like plain text is useful, for example with documentation or requirements. However, we all want to be able to "inject" nodes into this plain text, perhaps as a way of embedding references to other nodes. This will be possible with the Wiki Language.
- performance improvements
- potential Collections Language re-implementation using extensions methods to make it more light-weight
The goal is to introduce language versions and provide DSL users with the possibility to migrate their models from one language version into another while upgrading the language they use. In addition, we are going to record basic migration scripts automatically on any language modifications.
- Integration with Java profiler
- Detailed exception reporting that includes information about which language caused the problem
- Profiling particular DSL definition aspects like the editor creation process, completion menu and other
MPS refactorings will be improved in order to not only define custom refactorings for a particular language, but to contribute participants to existing refactorings from other languages. We will be able to record refactorings in order to replay them later on different set of sources.
The MPS generator can be used for transforming one model to another one
- expand set of target languages supported by MPS to Java Script/Objective C/...
- create public repository of MPS languages that would host IntelliJ IDEA language plugins
- implement new Base Language, which can be reduced into different target languages
- will better track cross-model dependencies better?
- will use “gps” builder to generate in the background during the editing
- will allow loading of XML schema/language grammar into MPS and generate custom persistence code for this language working in accordance with grammar/schema
Attach sources to Java libraries in MPS and navigate to the appropriate place in the Java source code with MPS debugger.
Eliminate the need for MPS models at runtime, so all you need to be able to work with your languages is Java source code.
Show history for single model root node (currently it's possible to see the history for whole model files only).