Zircon Zircomp 之类的东西与 ZeroC Ice 有什么不同?
目前有人在使用 Zircon Zircomp 吗?他们的技术与 ZeroC ICE 等技术有何不同?他们是否试图解决相同的问题领域?他们的目标受众有何不同?
附加问题:如果我可以在 Ice 和 CORBA 之间进行选择,根据您的经验,我应该选择 Ice 吗?
Is there anyone that's currently using Zircon Zircomp? How is their technology different from something like ZeroC ICE? Are they trying to solve the same problem domain? How's their target audience different?
Additional question: if I could choose between Ice and CORBA, should I go with Ice based on your experience?
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

发布评论
评论(2)
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
嗯,ICE 是一种简化的 CORBA,您仍然需要编写 idl、创建对象适配器并编写您自己的接口实现等。Zircomp 不需要任何这些。简而言之,它为您提供了一个使同步调用异步的工具,然后提供了一个可以运行任何此类调用/函数的“商品”服务器。 “商品”——因为你根本不需要修改它。唯一的编码是在客户端。您必须在库中提供原始(要分发的)函数,并提供函数参数库/标头位置的 xml 描述。请注意,库源代码本身也不是必需的。此类 xml 描述被提供给构建器,该构建器生成另一个要与客户端链接的库。它包含与原始函数具有相同参数的函数,但您现在在代码中使用该函数。该功能负责检测可用服务器、负载平衡、数据缓存、路由和丢失网络/引擎的恢复。回复通过屏障同步或回调到达应用程序。请注意,相同的功能可以配置为通过共置在本地执行,无需更改代码。在这种情况下,多线程/核心的使用将受益于没有复制或编组。
最后,没有 GPL 许可证,这是商业产品。
Well, ICE is sort of simplified CORBA, where you still have to write idl, create object adapters and write your own implementations of interfaces, etc. Zircomp doesn't require any of this. To put it simply it provides you a tool for making your synchronous calls asynchronous and then offers a "commodity" server that would run any such call/function. "Commodity" - because you do not have to modify it at all. The only coding is on client side. You have to supply your original (to be distributed) function in the library and provide an xml description of function parameters library/headers location. Note that the library source code itself is also not necessary. Such xml description is supplied to a builder that produces another library to be linked now with a client. It contains function with the same arguments as the original one, but that you now use in your code. And this function takes care of detecting available servers, load balancing, data caching, routing and recovery of lost network/engines. Replies come to the application via barrier synchronization or callback. Note, that the same functionality could be configured to execute locally, by co-location, no code change required. In this case multi-thread/core usage would benefit from absence of copy or marshalling.
And finally, there is no GPL license, this is commercial product.