Donald Ball wrote:
> oh, do we have the cocoon: protocol implemented now? oh, note that i still
> remain adamantly opposed to using the xinclude namespace if we don't
> conform (largely) to the xinclude specification...
I am not sure--I still have to double check. The change was made a couple
days ago, and I have been split three ways on projects at work for a while.
> ah, but the problem with using strictly sitemap transformers to do this,
> at least for webapps, is that you don't always know what transformers
> you're going to need to use in a given pipeline, and the pipeline is
> immutable. so suppose i have a transformer which operates on my namespace
> and transform
>
> <my:get-clients/>
>
> into sql transformer elements:
>
> <sql:connection>
> ...
> <sql:execute-query>
> <sql:query>select * from client_table order by name</sql:query>
> ...
> </sql:execute-query>
> </sql:connection>
>
> for this to work in the current environment, i'd have to known that the
> 'my' transformer is probably going to need to call the 'sql' transformer
> immediately after it. that should not be the concern of the sitemap
> maintainer...
And we are back to another one of your points on XSP: ordering is difficult
and has profound effects on the resultant code.
S/MIME Cryptographic Signature
