Sylvain Wallez wrote: > > Hi team, > > I just checked in Greg Weinger's AsynchronousCIncludeTransformer [1] in > the 2.0 scratchpad. > > This transformer allows included resources to be fetched in parallel > using concurrent threads, which is tremedously useful when included > resources are slow such as databases queries or remote web servers (e.g. > RSS feed aggregation). > > There's a sample under mount/asynchronous (Greg, I modified it a bit). > > However, this makes yet-another implementation of the include family, > and it's more than time for us to try to unify those. > > Should we stick to our own CInclude, or should we go back to XInclude > [2] which has reached its 2nd candidate recommendation last september > (how slow : the first draft was written nearly 3 years ago !) ? > > A quick look shows that XInclude allows attributes in other namespaces > on <xi:include>, which we may use to give some parameters to the > transformer for a particular resource, and also supports an > <xi:fallback> in case the included resource isn't available. > > Is this enough for our needs ? > Definitely no, the current CInclude transformer can do much more than the simple XInclude version. You can configure the http connection (set the timeout, set if you want to do a post etc.) and you can add parameters described in XML. So I think we really need the CInclude version - if we want to implement the XInclude standard, then we need two transformers. They can - as discussed many times - share the same implementation.
Carsten --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
