ipe | 项目脚手架 ExtJS4.2 Spring3 Hibernate4 | Object-Relational Mapping library

This question is indeed about different integration options. In general, it is one or more Application Entities (AE) that operate with the resources / data in a CSE. Every tool, application, protocol gateway, or "adaptor" is, from oneM2M perspective, an AE.

A special form of AE is the IPE, the Interworking Proxy Entity. It is used to connected another system or infrastructure to a CSE and mediate between these two worlds. The CSE part, ie. the protocol and the data formats etc, are covered by the oneM2M specification, and the other part must then implement the other protocol and technical specifications. The actual task is then to convert the data and expected behaviour between both worlds.

I am not sure from your description whether what you are planning is such a protocol "gateways", or just push the data received via an ftp server to a CSE. But technically, this doesn't matter much for the oneM2M part: You need to define how the data is structured on the CSE etc anyway.

For the data retrieval: As you mention in your question there is a third option to the other two:

1. The AE monitors an external system for new or updated resources, and pushes them to the CSE as soon as when they are ready.

From oneM2M's perspective this is completely fine. Technically you must make sure how to find out when the data is new, completely received, still valid, etc. ... basically all the challenges when you are polling resources. The first two options would be more helpful because you would avoid some of the problems that come with polling, but as you wrote sometimes you just don't have control about the system you want to connect to.

This is a question more related to the infrastructure architecture you are planing to deploy. It is feasible, of course, but polling has the disadvantage that if the delay is to small your AE would poll the ftp server unnecessarily often, and if it is to long then you might miss data delivered by your device (to the ftp server). Alternatives could be:

• The ftp server notifies your AE
• Your AE implements the ftp server and would then be able to directly react on new data events.

Your approach would work. Any AE that has the permissions to create a content instance under the container can set configuration data this way. The CONFIG-AE in your example would then need to apply the new configuration accordingly. I am not sure, though, why this AE would be an IPE? With what would it provide interworking functionality?

Nevertheless, you should also have a look at TS-0001, clause 10.2.8 "Device management" and the whole of TS-0022, "Field Device Configuration". Here, oneM2M specifies specific management resources to manage nodes in a oneM2M deployment. This might look like an overkill at first but since the resource types defined there are well aligned with other management technologies this might be worth the efforts.

Depending on your infrastructure you might also want to look at TS-0005 "Management Enablement (OMA)" and TS-0006 "Management Enablement (BBF)" in case you are working with remote management technologies from OMA or BBF.

I would use data-attributes instead of the JS object

I found the solution:

My css now is the following:

The solution is to replace with a group , while also moving overflow="visible" inside of the that belongs to the group.

I think the simplest approach is to offset the date column and sysdate by 3 months, and compare the year part:

Don't use Connect().

UDP is not a connection-oriented protocol in the first place. In spite of this, the Connect() method is supported for UDP sockets, as far as I can tell, simply for convenience. In simple scenarios, calling Connect() once is simpler than having to keep passing the remote endpoint address with each call, and it also has the effect of filtering inbound traffic. These can in some cases make the code easier.

But in your case, it's not making things easier. So don't use it. There are Send() and Receive() overloads that allow you to specify the remote endpoint for each send, and obtain the remote endpoint for each receive.

If you can use the Socket class, that class provides SendTo() and ReceiveFrom() methods for the same purpose. UdpClient is just a thin wrapper around Socket, so often that's an easier/preferable way to go anyway.

As long as you don't call Connect(), you should not run into the problems you're seeing.