Dmitry Maslennikov · Mar 4, 2016

Contributing to an Atelier

As I know Atelier sources already on github, but I think in private repository. May you say when this project will be opened, and how about if I want to contibute to this project?

0 314
Discussion (7)4
Log in or sign up to continue

This is a good question. One of the potential benefits of moving to Eclipse is the increased potential for an ecosystem compared to the Studio days. I think that even if Atelier source isn't opened up to us we will still be able to add stuff around it. But it would certainly be cool if InterSystems released it as open source.

We are still deep in a development phase. Robert Nagle does want to open source Atelier development at some point further along in the process. Its not yet clear exactly how we will do that but keep an eye here for an update.

There's where I find it a pity ISC choose Eclipse as a base for Atelier (love the name btw). Besides that I'm not a fan of Eclipse, I find it bloated and non-intuitive, the fact that plugins/extensions  need to be written in Java limits  (at least for me) the possibilities. A Javascript based IDE (such as Atom or Brackets) would make it much more accesible. That said, if ISC would be able to divise a plugin structure that is COS/Server based (at least partially) that might open up a lot of opportunities for seasoned COS developers.

Herman, there is REST API to work with COS codebase in Caché which can help to implement client side in any IDE or editor you like e.g. Atom, Brackets or Sublime.

Well, but maybe possible to get COS parser as a separate project. What I see now, that most important errors in Atelier for me now, it is Parsing errors. And such problem not only for Atelier, for any other editors which now available.  And parser as a one of the most difficult parts of editor should be done by InterSystems, they change language. While InterSystems have already opened server's  API, as a chance for any others developers to make their own editor. Next step as I think should be open Parser, or any help in support. I know exactly how difficult to write COS parser, I tried to do it for my NBStudio, and helped in writing SonarQube plugin, and last one my project it is a pygments COS parser used for GitHub. 

And finally I think, to get most support for all future editors, Java-based or JavaScript or even C#, such parser should be written on ANTLR. Now it supports not only java as a platform, JavaScript and C# supported too, and I think this parser would be used for any future editors, and Caché-developers finally gets more then one useful editor.

The SQL parser, the HTML/CSP parser, and the JavaScript are all ANTLR.

The Atelier project and its COS parser were started before the strengths of ANTLR4 were realized.

I agree that a great strength of ANTLR4 is that once one has written the grammars, as long as those grammars don't involve too much code, it is trivial to generate a parser in the language of one's choice whether JavaScript, C#, or Java.

I think (personal opinion) it would be good to have on-going discussions of an ANTLR approach to parsing COS.

Generally we favor ANTLR. The COS parser required complexity we were not confident we could achieve with ANTLR. Having said that we continue to debate this internally as Jonathan expressed. When we release Atelier we do anticipate publishing a lot of the code to open source. Dmitri I will reach out to you through the Russian office as we may be able to leverage your knowledge.