The OpenSearch protocol developed as part of Amazon’s A9 has been pretty much dismissed by the SRW community as just being too limited to be of interest. The idea of SRW was to make the functionality of Z39.50 available in a web-friendly manner. Evidently A9 didn’t think it was quite friendly enough, and has come out with a simpler protocol, quite cleverly built on top of RSS, for easy access via news readers.
To me, however, OpenSearch looks like it might be very appealing to something like the NISO Metasearch Initiative, which just had a meeting in Chapel Hill. OpenSearch has the feeling of the sort of thing where 'simple wins' and those with more complex technologies ignore it at their peril.
Is there a middle ground here? Something that is compatible with SRU, but closer to A9? Our Metasearch representative, Ralph LeVan thinks so and has offered to develop guidelines to make this happen. The catch is that what is being proposed, while compatible with SRU, is not SRU compliant. According to Ralph, there is some reluctance on the part of those developing SRW to embrace anything less than full compliance, but we argue that not only would the proposed simplification of SRU be a step forward from the screen-scraping that Metasearchers engage in, it is a step in the right direction, in that extending the services to full SRW compliance wouldn’t break anything.
Here are two from the A9 search query page:
To me those two examples look incompatible, but maybe the '?s' in the first should have been '?q'.
Here's an SRU search:
The SRU example looks simple enough, although it won’t come back as RSS automatically.
I’ve argued in the past (with only moderate success) that if you are doing something that is similar to an OpenURL request you really should put it in the OpenURL syntax, even if you aren’t bringing up a full OpenURL server. It gives you a natural path for future development, avoids incompatibilities and you don’t have to spend time developing yet-another funky syntax.