ICE std-ization?

marlowa

I would like ZeroC to consider std-izing ICE at some point. I think ICE is a great improvement on CORBA but without std-ization it could be perceived as a high risk strategic technology choice. I realise that the risk is mitigated by the fact that it is open source but even so....

Std-ization would, of course, open up the field for competing proprietary implementations. I realise that ZeroC might not be keen on that. But CORBA vendors seem to be surviving by implementing an open std. Iona is the leader but that is probably mainly due to theirs being the first implementation to hit the streets. Similarly, if ICE was std-ized then the ZeroC implementation would be dominant by being there first, at least that's my theory FWIW.

marc

Actually, we would welcome competing Ice implementations! First, we believe that competition is a good thing in general. Second, we all have big egos here, and therefore we would love to see others to also implement the technology we invented

Let me point out again that we don't hold any patents in Ice, so anyone can take our Ice documentation and write a competing implementation. What they cannot do, of course, is to use the source code of our Ice implementation, i.e., it would have to be a clean-room implementation. (Except if this implementation would also be released under the GNU license.)

marlowa

marc wrote:

we don't hold any patents in Ice, so anyone can take our Ice documentation and write a competing implementation. What they cannot do, of course, is to use the source code of our Ice implementation, i.e., it would have to be a clean-room implementation. (Except if this implementation would also be released under the GNU license.)

Thanks for the clarification. But actually what I was asking about was std-ization. Open source means competing implementations can arise but they cannot be de jure stds without some sort of std-ization process.

michi

Standards exist to prevent implementations from different manufacturs from deviating from each other. As long as there is only one implementation of Ice, there is no point in standardizing anything -- the ZeroC implementation is the standard. And standards gain credibility by the organizations that support them, and the organizations that publish them. A standard that is written by ZeroC and published by ZeroC, with no other manufacturers and standards organization involved, would be meaningless.

If standardization is ever to happen for Ice, we first need implementations from a source other than ZeroC, and then we can think about standardizing things.

Cheers,

Michi.