Hi Thomas,

I can see breaking compatibility on a module basis: in fact this is what I suggested to DinoV. It turns out there had been quite a discussion about this. At one end are the purists, who want compatibility never to be broken. At the other end are the pragmatists, who just want a clean, dynamic language that returns .Net-typed objects. I think the middle position makes sense. Boo does for duck typing either on an object basis (by implementing IQuackFu) or on a module basis (by compiling with -ducky). I think the middle ground that allows the user to work either way doesn't detract from the experience of either kind of user.

The IPy team is, I think, 2 or 3 people. And Jimmy did a lot of work on IPy. Hardly reassuring, even given that they are extremely talented.

Hank

>Hi Hank,
>>
>> It turns out the decision was made before the announcement, which mirrors the amount of trust I have in MS as the keeper of the languages in which I develop.
>
>Uurgh, that is even worse. Have not researched the manstrength of the IPy team: any changes in that area visible as well ?(I realize that most of the dynamic team had to wear different hats part of the time, which is not really helpful in such calculations...)
>
>>Yes, IPy is a faithful implementation. For me that is a negative, not a positive. Languages need to grow to meet the needs of new contexts, and .Net is a new context.
>
>As I had the chance to play with Jython when doing some java work and small paying jobs in CPython this is an area I disagree - compatibility gave me the opportunity to use it. IMHO platform differences/extensions should be handled in modules. Aspect oriented programming at the compilation process should be handled at the language level, keeping the modules clean to at least compile in each environment and fail if the needed services are unavailable. And I had the impression that there was some exchange between CPy and IPy about possible extensions - so I guess it is not as one-sided a process as it seems to all wanting enhancements.
>
>>As to how I'm taking it: open source projects are credible to the degree they are corporate-supported. Boo is supported (Rodrigo works for Unity, the game development framework company, and Boo is the basis for their scripting language, and all Unity developers have 20% time for non-Unity development).
>
>This works great as long as there are no wishes to fork/fragment. Remember the db3+/db4/Clipper/dBMan/Quicksilver stage of the xBase language...
>
>>And if VFP.Net were to be resurrected, I could get multi-company support for it, probably within 24 hours. I prefer to put eggs in a basket in which I have a reasonable degree of trust, and with regard to development tools, MS does not meet that criterion, except with regard to languages that don't meet my needs.
>
>At least you have the code to try to fix if a specific behaviour is causing trouble.
>As for enhancing/keeping current, you are probably correct.
>
>regards
>
>thomas