Subject: RE: The Cathedral and the Bizarre (was: do you use pi's?) From: "Didier PH Martin" <martind@xxxxxxxxxxxxx> Date: Wed, 7 Apr 1999 11:59:40 -0400 |
Hi Guy, We have first to agree on a mechanism for MIME filters. We are documenting a XPCOM interface for this. I guess we will have something working after M5 or after the 5.0 release. Actually, we have code freeze and now its time for bug resolution, so its not time to change anything. But after the release we'll implement and document the MIME filter concept for the HTTP protocol handler. Mozilla development process is not as fast as we would it be, but we now progressing faster than before. The problem is that it is still a Netscape owned project and there is no documentation. the learning curve for a newcomer is big. You have to learn form source code. However, the new version is an enormous improvement on the last. Everything is now formatted as modules. There is a binary signature convention so that a module can talk to an other one on the same platform. You could easily create a COM/XPCOM component with actual tools with minor modifications. If we can get people to document this, developers will discover some pearls there. The layout engine is getting stable from day to day. There work to do on the next version: a) plug and play for XPCOM parsers. A parser has access to the DOM and do the translation from a format to the DOM. Quite powerful. Potentially you can create a parser for the language of your choice and not necessarily XML or HTML. Or implement DSL (Domain specific language) extension to existing languages like for instance XLL , Hytime or XPointer. b) NetLib MIME filters that could be liked together in a transformation chain. For example, a filter could decompress a document received form a HTTP 1.1 server (with compression ON), an other one could de-encrypt, an other one could transform the document with a style sheet and finally the result either a) parsed and displayed, b) is a DOM tree created by the style engine and then given to the layout manager for display. Exiting mechanisms, What now needed is that the development community become aware of these good things. So, to now answer directly to your question. Do not expect something working for XSL, DSSSL or XScripts before this fall. I we are lucky and get ahead of schedule this could be before, but I got burned more than one time with slipping schedules in this project :-))). But the actual code freeze do includes CSS and XML processing. However (just to put pizzaz on the actual thread on validation) the parser is a non validating XML parser. Guys from DocZilla created their own version with their SGML/XML parser. This is a validating parser (obviously because it is a SGML parser that can also parse XML). This could have been a solution but we want to sit the interfaces in the specs so that anybody could add new features and get a high probability to have this stuff work. This inlcudes being able to replace the parser. So, if you don't laike the XML or the HTML parser, you can replace it with your own. Validation "is a must" supporters will then be able to replace the provided HTML/XML parsers with their own parser as long as it is XPCOM compliant and fulfill the interface contract. But stay tuned, we'll have something new very soon. A procedural event base script language named XScripts. You'll be able to transform a SGML/XML document into whatever you want (not restricted to XML). Here is a sample <XStyles> <element match="/"> output"<HTML>" process-children() output"</HTML>" </element> <element match="MyMarkup"> output "<BODY>" output "<P> This text includes" & (Sheet * 2) & "pages<P>" output value-of_select("subMarkup") </element> </XStyles> The body of each rule could be in a) javaScript, b) VBSCript, c) PERLSCript, d) PythonScript. So its a mix of rule based and procedural programming. This will work with Mozilla and IE and could be attached to a XML/SGML document as a style sheet or use to process documents in batch. Its simple, as you see, the rules are fired when a match is encountered. We are studying if XQL queries could be used for the match so that these two universe could be merged (it makes sense that they should). We also, learned from XSL and DSSSL and kept several concept from both languages but adapted this to procedural languages. We kept the concept of rule. Actually, the style document is a SGML document. if someone want that the style script is a "kosher" XML document then a rule as to be enclosed with the CDATA[....] construct. Note that all actul script constructs, numerical and string processing can work. Same thing for object creation and usage. You can for instance create a data base object and use it in a rule. It is simply versatile. Now we try to minimize the Script to Grove/DOM interface like people in Perl-XML experimented. We also learned a lot from their experimentation with expat-XML interfacing. The concept of style router is making progress and I'll publish something on this very soon. Again, we are learning a lot from the work we did on the SGML/XML kit that routes to XSL, CSS and DSSSL style sheets. By the way, we'll post a new version based on a different module packaging. We now know more about the pros and cons of MIME filters and document handlers. I'll keep you informed Guy if you want to make some tests with XScripts. Your pragmatic mind and feedbacks are more than precious. Regards Didier PH Martin mailto:martind@xxxxxxxxxxxxx http://www.netfolder.com -----Original Message----- From: owner-xsl-list@xxxxxxxxxxxxxxxx [mailto:owner-xsl-list@xxxxxxxxxxxxxxxx]On Behalf Of Guy_Murphy@xxxxxxxxxx Sent: Wednesday, April 07, 1999 10:07 AM To: xsl-list@xxxxxxxxxxxxxxxx Subject: RE: The Cathedral and the Bizarre (was: do you use pi's?) Hi Martin. When do you envisage having something that people can play with? Cheers Guy. xsl-list@xxxxxxxxxxxxxxxx on 04/07/99 04:30:09 PM To: xsl-list@xxxxxxxxxxxxxxxx cc: (bcc: Guy Murphy/UK/MAID) Subject: RE: The Cathedral and the Bizarre (was: do you use pi's?) Hi Guy, We are working on that. There is actually two solution envisioned: a) a MIME filter is inserted in the protocol handler (similar architecture to IE 5 but this time, not with an unknown interface :-) b) a DOM to DOM transformation. However this last process requires XSL formatting object to be implemented. We are looking at both solutions, but the first one is actually the easiest one. regards Didier PH Martin mailto:martind@xxxxxxxxxxxxx http://www.netfolder.com [SNIP] XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
RE: The Cathedral and the Bizarre (, Guy_Murphy | Thread | RE: The Cathedral and the Bizarre (, Guy_Murphy |
Re: XPDL (was Re: XML is broken), Simon St.Laurent | Date | Re: XPDL (was Re: XML is broken), Guy_Murphy |
Month |