Hi, Here is a short update of what we have covered today in the Chemistry face-to-face meeting.
* Presentation of OpenCMIS [1] and Chemistry [2] concepts and ideas. Getting everyone on the same page. No news, just a summary of what has been described on the Chemistry Wiki and on the mailing list. * Moving forward with Florents merge proposal. We use Florents merge proposal [3] as the basis for the next discussions. * Discussion about the low-level client API and server SPI. Chemistry has the same set of interfaces for both, the client API and server SPI. That allows to set up a client in the same JVM as the repository without a network connection. (See [2] for details.) Unfortunately, the current approach cannot provide a fully complaint AtomPub implementation. OpenCMIS has separate interfaces for the client and the server. The client API is modeled after the CMIS specification and independent of the binding. The server SPI takes binding details into account which is necessary to provide complaint AtomPub and Web Services interfaces. We all agree that a local loopback would be useful. The discussion on how to achieve this hasn't been finished. There are two basic proposals: 1. Keep the different client API and server SPI and build a thin layer that does the translation. 2. Rework the API and SPI that they eventually become one. There is probably no way do this without introducing operations and/or parameters that don't make sense for users of the client API. We will dive deeper into this discussion tomorrow. - Florian [1] http://people.apache.org/~fmui/opencmis.pptx [2] http://blogs.nuxeo.com/fguillaume/2009/06/promises-modern-chemistry.html [3] http://mail-archives.apache.org/mod_mbox/incubator-chemistry-dev/201004.mbox/%[email protected]%3e
