Hello Graham, Thursday, February 02, 2006, 1:15:44 PM, you wrote:
GK> So is there a compelling feature in this Eclipse plugin that isn't easily GK> achieved using simpler tools?
it's a list of what-i-want-from-the-IDE:
* autogeneration of prototypes GK> I don't know what you mean by this.
adding type signatures of functions to module. this can be done by loading module into the hugs/ghci and using ":t"; or by the programmatica tools, afair
* context help GK> Very nice - might persuade me to use Eclipse routinely (I assume you mean things GK> like getting help about functions, esp. prelude and standard libraries)
yes. btw, if you need it, you can use hasktags with a vim/emacs. it allows you to jump right to the fucntion definition, which is more informative and anyway typically contains text for the haddock. vim even can show this function definition in the supplementary window without jumping to it
* type-based function hierarchy browsing GK> Very nice. I assume this allows functions to be located from type signatures, GK> including prelude and standard libraries.
seems that we say about different things. IDEs for OOP languages show program hierarchy around the defined classes, that is unuseful for Haskell. nevertheless, many functions typically grouped around types they "support". so, a "module->type->function" view and "type->function" view should be handy GK> In summary, I think there's enough there to make me think seriously about using GK> Eclipse for Haskell development, at least some of the time. sorry, i said about ideal IDE, not about Eclipse (what i never seen, after all). and of course i'm not the author of this package. sorry :))) GK> Something I'd also like is in-editor warnings of features that are not supported GK> uniformly across different implementations -- my (rough) criteria for "portable GK> Haskell" has been to check that code runs with GHC and Hugs. one thing that i skipped in my dream list is on-the-fly syntax checking GK> Also, some form of integrated testing support would be nice. There was brief GK> discussion here some time ago about something like assert statements that could GK> be checked at compile time -- if you have in-context expression evaluation then GK> this wouldn't be a great leap: e.g. scan the code for identifiable expressions GK> that are expected to be true, and report warnings if they are not. good idea. i think that i should place all our stuff at the "IDE/Ideal" wiki -- Best regards, Bulat mailto:bulatz@HotPOP.com