The current trend is to make MPS less dependent from IDEA.
This doesn't mean that we are going to include java tools in MPS.
Thus, if you are developing a language in MPS you still need to run some external java tool to compile, debug etc.
Accordingly, if you do something that targets non-java platform you will need to run other external tool.
Regarding language development in MPS, the major reason why you need integration with IDEA is ability to create/navigate-to so-called 'query methods' - these pieces of java code which cover functionality that we can't express using existing DSLs.
We are going to get rid of the 'query methods'. For instance, in next MPS build some of editor actions can be written completely in dsl.
In future, we also going to get rid of generation of java-api for the language structure.
I think, MPS is a very interesting
project, but it appears the progress stalls and there
never will be a point,
where we can use it for real-world projects
No, the MPS doesn't stall, as general purpose LOP-oriented tool it is and will become better and better over time.
Regarding real-time projects, so far we successfully use MPS to develop itself but we are trying to apply LOP to domains other then the MPS development.
When we succeeded we will certainly share the methodology and tools with everybody.
For instance, recent moves in the field of semantic web seems very promising:
I'd like to ask again: is there some roadmap available?
Tom
Hi, new MPS build may come out in 2-3 weeks.
It will include bugfixes, usability improvements and more major reworks in our structured editor.
I expect updated 'action system' will be more extensible and customizable.
Btw, current 'action sets' now considered deprecated.
Igor.
Igor, no problems with waiting. But my main problem with MPS (and some other
Jetbrains tools) is the lack of information or communication at all. Into
what direction MPS is heading? What features the version 1.0 most likely
will contain and for what the user will be able to use it in production
environments? When it is planned to be released (quarter of year would be
enough)? It's not that amusing to see Jetbrains invest a lot of effort into
Fabrique or MPS and finally one or another project might be stalled.
Thanks in advance,
Tom
Hi Tom,
I may dissapoint you, but we don't have roadmap like "feature f1,f2,.. by date1 ; f3,f4... by date2" and so on.
Our development strategy (at least now) is not based on the MPS v1.0 release as target.
Instead, we are constantly trying to apply our in-hand technology on practice. Then we can better see flaws in MPS and fix them immediately.
In my view, we will rather release some cool domain-specific staff then MPS v1.0.
Igor.
Igor, do you have a plan to either integrate MPS into IDEA or make it
independent from IDEA? It's quite annoying to have both applications open.
Please don't understand me wrong. I think, MPS is a very interesting
project, but it appears the progress stalls and there never will be a point,
where we can use it for real-world projects. This would be very
disappointing, because IMHO MPS or an MPS-like application has very much
potential. I'd like to use it rather today than tomorrow, but also need
support for it.
Tom
The current trend is to make MPS less dependent from IDEA.
This doesn't mean that we are going to include java tools in MPS.
Thus, if you are developing a language in MPS you still need to run some external java tool to compile, debug etc.
Accordingly, if you do something that targets non-java platform you will need to run other external tool.
Regarding language development in MPS, the major reason why you need integration with IDEA is ability to create/navigate-to so-called 'query methods' - these pieces of java code which cover functionality that we can't express using existing DSLs.
We are going to get rid of the 'query methods'. For instance, in next MPS build some of editor actions can be written completely in dsl.
In future, we also going to get rid of generation of java-api for the language structure.
No, the MPS doesn't stall, as general purpose LOP-oriented tool it is and will become better and better over time.
Regarding real-time projects, so far we successfully use MPS to develop itself but we are trying to apply LOP to domains other then the MPS development.
When we succeeded we will certainly share the methodology and tools with everybody.
For instance, recent moves in the field of semantic web seems very promising:
http://redherring.com/Article.aspx?a=14902&hed=SemanticWeb%2cHereWeCome
There will be demand for tools capable to edit structured data directly, to manage ontologies and corresponding services.
Igor.