You are viewing a plain text version of this content. The canonical link for it is here.
Posted to slide-dev@jakarta.apache.org by dog <do...@dog.net.uk> on 2000/12/14 13:21:55 UTC
introduction and issues
good morning slide-dev.
i work with a research team in london, and we'd like to help with work on
slide with the aim of using it for the basis of a content management system.
initially, we'd like to work with you to define the apis for the
index/search helper and the process helper. we have considerable experience
in both these fields.
i have a few issues i'd like to discuss to start with:
1. xsl transformations.
let's say i have a namespace consisting largely of xml documents, xsl
stylesheets, and schemata, and i would like to process the xml with the
associated xslt resource as part of a streamlined delivery. at the moment it
seems i could go about this in one of two ways:
a. use a servlet/ejb/jsp to manage the transaction, invoking the stylesheet
processor on the content retrieved from slide.
b. use servlet chaining such that a stylesheet processor (eg cocoon)
postprocesses the output of the dav servlet.
what are the performance implications of these options? has anyone tried to
do this yet?
2. input validation, datatyping, etc
in more or less the reverse of the above, i'd like to check that incoming
xml content is valid according to its schema (dtd or xsd). this process
should probably be related to indexing functionality. assuming that the
schema exists in the slide namespace, how would i go about this? do you feel
it should be part of normal slide functionality, a separate servlet, or
another helper api? has anyone considered the role of digitally signed
and/or encrypted content with respect to this kind of content preprocessing?
3. search
is dasl on the to-do list for slide? it would seem to be the way to go as
search languages can just be plugged in.
4. indexing
what are people currently considering when they're talking about indexing
for slide? just text indexing? i'd like to index xml such that any xlinks in
the content are automatically updated when their target moves, for instance;
presumably this interfaces with the structure helper, but slide is pretty
content-type-agnostic. again in terms of preprocessing: could we selectively
register content-type handlers with dav/structure/content methods? this
would tie in with issue 2 above.
5. process
at a minimum we would want to register a process engine with slide to be
informed of clock and user events. is there anything else that people feel
strongly needs to form part of a process api?
tia
--
dog
Re: introduction and issues
Posted by dog <do...@dog.net.uk>.
sculpsit daveb:
> > is dasl on the to-do list for slide? it would seem to be the way to go as
> > search languages can just be plugged in.
>
> Yes. Do you guys have any experience with natural language searches :-)?
no. we do have some experience of natural language parsing, if that helps :-)
> > what are people currently considering when they're talking about indexing
> > for slide? just text indexing?
>
> For the indexing, we've discussed making them pluggable in the same fashion
> as the Descriptor/Content stores and configurable via the Domain.xml file.
> My thoughts, as a start, were to use lucene ( www.lucene.com ) as the
> indexing engine and to initially add support to search the webdav
> structures. Content type handlers could be added separately as plugins.
sounds reasonable. i'll have a look at the lucene engine and see what it
needs.
--
dog
Re: introduction and issues
Posted by Dave Bryson <da...@miceda-data.com>.
On Thu, 14 Dec 2000, you wrote:
> good morning slide-dev.
> 3. search
>
> is dasl on the to-do list for slide? it would seem to be the way to go as
> search languages can just be plugged in.
Yes. Do you guys have any experience with natural language searches :-)?
> 4. indexing
>
> what are people currently considering when they're talking about indexing
> for slide? just text indexing?
For the indexing, we've discussed making them pluggable in the same fashion
as the Descriptor/Content stores and configurable via the Domain.xml file.
My thoughts, as a start, were to use lucene ( www.lucene.com ) as the
indexing engine and to initially add support to search the webdav
structures. Content type handlers could be added separately as plugins.
--
Dave Bryson
daveb@miceda-data.com
----------------------
Re: introduction and issues
Posted by dog <do...@dog.net.uk>.
sculpsit remm:
> > 5. process
> >
> > at a minimum we would want to register a process engine with slide to be
> > informed of clock and user events. is there anything else that people feel
> > strongly needs to form part of a process api?
>
> Slide right now is very centralized and does everything synchronously. While
> you can probably plug-in a process engine, but it will have to do very
> simple processes (no human interaction, ...).
> I'll eventually propose a SlideNext which will probably be based around a
> specialized process engine, and I'll try to make it more distributed.
> However, that won't happen in the near future (obviously, I'm concentrating
> on TC 4.0 and Slide 1.0 right now ;).
we plan to write a process engine that will support the workflow management
coalition's standard interfaces, eg wf-xml
http://www.aiim.org/wfmc/standards/docs/Wf-XML-1.0.pdf
we hope to collaborate with you to make slide a perfect content delivery
mechanism for this process engine.
have a good holiday and we'll talk further in the new year.
--
dog
Re: introduction and issues
Posted by Remy Maucherat <re...@apache.org>.
> good morning slide-dev.
>
> i work with a research team in london, and we'd like to help with work on
> slide with the aim of using it for the basis of a content management
system.
> initially, we'd like to work with you to define the apis for the
> index/search helper and the process helper. we have considerable
experience
> in both these fields.
>
> i have a few issues i'd like to discuss to start with:
>
> 1. xsl transformations.
>
> let's say i have a namespace consisting largely of xml documents, xsl
> stylesheets, and schemata, and i would like to process the xml with the
> associated xslt resource as part of a streamlined delivery. at the moment
it
> seems i could go about this in one of two ways:
>
> a. use a servlet/ejb/jsp to manage the transaction, invoking the
stylesheet
> processor on the content retrieved from slide.
>
> b. use servlet chaining such that a stylesheet processor (eg cocoon)
> postprocesses the output of the dav servlet.
>
> what are the performance implications of these options? has anyone tried
to
> do this yet?
The best way to do that would be to provide an interface for a presentation
engine like Cocoon, which would be way more fleible than anything you could
come up with.
> 2. input validation, datatyping, etc
>
> in more or less the reverse of the above, i'd like to check that incoming
> xml content is valid according to its schema (dtd or xsd). this process
> should probably be related to indexing functionality. assuming that the
> schema exists in the slide namespace, how would i go about this? do you
feel
> it should be part of normal slide functionality, a separate servlet, or
> another helper api? has anyone considered the role of digitally signed
> and/or encrypted content with respect to this kind of content
preprocessing?
I think it should be a process.
Since there's no process manager yet, you can do the validation as part of
the Store.
> 3. search
>
> is dasl on the to-do list for slide? it would seem to be the way to go as
> search languages can just be plugged in.
Yes. However, being able to plug a random language could be hard, since the
searching engine (which sits way under the DAV layer) would probably also
have to support that language.
> 4. indexing
>
> what are people currently considering when they're talking about indexing
> for slide? just text indexing? i'd like to index xml such that any xlinks
in
> the content are automatically updated when their target moves, for
instance;
> presumably this interfaces with the structure helper, but slide is pretty
> content-type-agnostic. again in terms of preprocessing: could we
selectively
> register content-type handlers with dav/structure/content methods? this
> would tie in with issue 2 above.
The searching / indexing will be done like all the other modules, as Dave
mentioned.
Since we have metadata, we'll only require the indexing engine to index the
metadata (that's the whole point, IMO; otherwise, why have metadata in the
first place), and optionally index the content too (using an appropriate
parameter, like the content type, to decide whether or not that should be
done). DASL is mostly metadata searching oriented.
I totally agree on choosing Lucene as the default index store.
> 5. process
>
> at a minimum we would want to register a process engine with slide to be
> informed of clock and user events. is there anything else that people feel
> strongly needs to form part of a process api?
Slide right now is very centralized and does everything synchronously. While
you can probably plug-in a process engine, but it will have to do very
simple processes (no human interaction, ...).
I'll eventually propose a SlideNext which will probably be based around a
specialized process engine, and I'll try to make it more distributed.
However, that won't happen in the near future (obviously, I'm concentrating
on TC 4.0 and Slide 1.0 right now ;).
Remy