Re: [ide-dev] A story of Handly and MelnormeEclipse
I'd rather keep Handly separate for the time being, with perhaps moving parts of it down to the Eclipse Platform (LTK?) eventually. I believe that it is not necessary to be part of another project to be used by it, right? :-) I mean that DLTK could depend on Handly, not the other way around. (Handly is really much more generic while being much narrower in scope.)
In theory yes, in practice, there are other things to consider than just code:
* Handly is currently a small and young project: very few contributors, not much community... Those are things that need to tremendously grow to consider Platform or DLTK relying on Handly. Indeed, adopting an upstream project as a dependency is a critical strategic choice, so it requires real sustainability to be considered. However, in the other way round, if Handly gets part of DLTK, then it will immediately get a boost of contributors and/or users, and be perceived as more "reliable" by other projects such as JDT, CDT, JSDT, PDT...
Sorry, but I don't see how this boost can magically happen without at least DLTK adopting Handly right in their core first. And if such adoption really happens, I think that Handly will immediately get a boost of contributors and/or users as you suggested, regadless of whether it's part of DLTK or not.
It is not clear to me what exclusive benefits DLTK would receive from "hosting" Handly? If someone is interested in becoming a committer on Handly, there is a well-established and quite easy to follow process for that.
Handly was, is, and will remain for foreseeable future an active project open to quality feedback and contributions from whatever interested parties. It is backed by a major adopter that uses Handly in its enterprise solution. I don't quite understand how being part of another project that itself doesn't have an exceptional history of reliability would make Handly seem immediately more reliable.
Speaking of reliability, you can consider the fact that, despite being quite heavily used in a couple of commercial IDEs for some time and being in version 0.4 currently, the number of "enhancements" in Handly, as evidenced by Bugzilla, is vastly greater than the number of "bugs", with the rate of "bugs" steadily declining with time. In particular, I don't remember we have had such a thing as NPE (I'm pretty sure we'll have one some time :-) ).
* Moving a piece of code to Platform or JDT comes with a bunch of constraints about API or release schedule, so it's clearly not something recommended for any piece of code that didn't reach a high enough level of maturity.
If you want to succeed in your goal of factorizing language tool, that cannot be a one-man project. It seems easier to join an established project that already share parts of this objective.
I'm well aware of that (that's why I said "perhaps" and "eventually"), and we really aim at Platform API quality (that's why we have not made a 1.0 release), but I still don't think that DLTK is the right home for Handly, sorry. There are substantial differences between the projects (goals, scope, approaches) in my view to keep them separate.
Anyway, thanks for your opinion! I appreciate it.