i found this article http://docs.google.com/Doc?id=dcsq25m_04k4cmgfd . generate html on server side for spider. what do you folks think about using this technique to serve different content depending on user-agent ?
any reference articles, u folks wanna share on best way to do seo on gwt? also, i did read up pro gwt2.0 technique, but seem to me it bit overkilled . hope to get feed back from anyone
The main problem is that search engines don't like it when you use different output when crawled vs visited "in person" by a browser, unless you provide a link on your app in the tag to the said html generated page.
also, the performance problems of using GWTTestcase to render html could be catastrophic given that it is not tuned for performance (but for testing and debugging) - in fact, crawlers tend to hit sites more than users do in a short time, and you might find that your CPU gets maxed out by search engines.
So far, there hasn't been any satisfactory solution to this problem unfortunately.
Hello,
We also had this problem developing joobili.com Honestly it was not easy, but I think the solution we have now is good.
We don't serve different content to search engines, because that is called cloaking, and google does not like it.
For every page we have an html version (honestly that is not a big deal, you don't have to deal with design at all, just put some content and links on the page).
You can use the tag for displaying this content. You can also speed up thing a bit, if you provide the initial content also in object form on the page. For that we use a slightly modified version of the kiyaa frameworks object serializer.
We also use gwt 2.0-s codesplitting mechanism, and this makes the loading of the homepage a lot faster.
Of course you have to make sure that the content you provide for the search engines (browsers without javascript) correlates to the one you provide to users.
Best - Istvan - inepex.com