Where is MPS heading? Follow
Although MPS seems like an interesting project, I cant seem to figure out where it is heading to.
Do you actually have sort of a milestone or release date in mind?
Also although the concepts of easily being able to create DSLs is interesting, how many people do you think really want to do that?
So although I find this project to be interesting, I dont see much real world usage or a market for it.
I would love to see someone from Jetbrains shed some light on this.
Please sign in to leave a comment.
Hello, Prashant. Nice to see you again in our forum.
Yes, we have a condition for MPS release. As soon as we have a really useful application created with MPS, we will release MPS. It's our current goal.
Now we trying to apply MPS to domains where it can be useful, for example we have a knowledge representation language which is included in our distribution. Recently we created a typesystem language. If you download last build, you can find its description in help directory. Probably in the next build several new languages will be included.
Everything goes on well. We keep trying to apply MPS in our internal software development.
Recently we doubled our team. New guys quickly caught up LOP ideas and look very enthusiastic toward the MPS (at least this one next to my desk looks so:)).
We've also created new detached project targeting web app development (DSL's for web app) and I can say I've already seen some coool thing out there.
Such an out of the box workbench would be fine milestone. We don't have release date set, sorry.
I guess most of them will be same people who currently develop frameworks+APIs in a traditional way. It is turned out that fundamentally there is no big difference with DSL's development. Almost every DSL comprises some internal part (platform?framework? which does a real work) and nice looking and savvy domain specific face (that often identified with DSL).
perhaps I am not completely understanding the concept, but it seems similar to what some of us who have been working with Object Oriented development for a while would call declarative programming. In other words: create objects that can be plugged together including those objects that represent relationships between other objects. This can be very useful particularly in GUI development where widgets are connected to objects (a la Visual Works Smalltalk using graphs of aspect adaptors) and where one can express the entire GUI using only creation of objects similar to dependency injection. One way of codifying the approach would be to call it a small language... Am I on the right track?
GUI in general is a domain (yet quite fuzzy) which can be farther divided into more concrete sub-domains like Web-GUI/Swing-GUI/SWT-GUI or MailClient-GUI/OnlineStore-GUI etc.
Going this way we most likely would come to a set of concepts (atomic units of a domain) which would be encoded in a DSL.
Probably, such a DSL will be mostly declarative. I believe that declarative languages are simpler and less error prone.
The example of GUI DSL is MPS's editorLanguage - we use it to define how a language concepts are shown in MPS node editor.