Subject: About new functionalities for OpenJade From: "Didier PH Martin" <martind@xxxxxxxxxxxxx> Date: Wed, 11 Aug 1999 21:54:44 -0400 |
Hi, Some of you may remember that I was working on a particular thread: have a XML/SGML document used as an input document for DSSSL processing and then have form this single document multiple outputs. More particularly to have the following behavior: the SGML/XML document contains arbitrary constructs (i.e. elements), these constructs are associated to certain DSSSL rules that: a) creates new groves and associated flow object tree. For instance, the usage of such system would be to create a batch file to process a collection of files. I added to the reflection some new constraints from comments made by some users in this list: a) there is a need to aggregate several input files for a single output file b) a need to create from a single file several output files Also, during my research on batch files, I came to the conclusion that each file referred in the batch processing document (expressed as a SGML/XML document) need probably to be associated to its own DSSSL script. So for instance, the following document may be used as a batch file: <?xml version="1.0"?> <batch> <process file="usr/dir1/dir2/mydoc.xml" script="usr/dir1/dir2/mydoc.dsl"/> etc... </batch> In the case where XML files respect the last W3C specifications and include a processing instruction, the source file - script file is made with this processing instruction. However, there is a need that this processing instruction is overridden or that no processing instruction is present and therefore some association has to be made. The above script may override or create the document-script association. What is required form the DSSSL script environment is: a) dynamically create a new grove and an new output file b) properly garbage collect the grove. In the case where a batch processing file is used, then: a) this file is transformed into a grove and processed by a DSSSL script. b) the script creates new groves - in order no to tax too much resources, groves should be garbage collected because, in this case, there is no intention to aggregate or do more processing when the root (process-children) is finished. c) In the case of page oriented output, some modifications have to be done on the simple-page-sequence to allow the creation of a new output file. I came to the conclusion that to implement this in OpenJade: a) a grove is released or garbage collected when the root element (implicitly or explicitly) process-children recursive processing is completed. When this occurs, the grove is released. b) a new characteristic has to be added to the simple-page-sequence to allow the creating of a new output file. If both conditions are met, then a fully transportable batch processing script is possible. SGML or XML based domain language created for batch processing could be anything we want, as long as the OpenJade environment provides conditions (a) grove released when the root (process-children) is accomplished and (b) a new characteristic is added to page oriented flow objects like the simple-page-sequence. There is still one issue, I have no mechanisms to dynamically change the script when several documents are processed. Actually, only one script (and externally included scripts) can process the collection of documents. I did not found any mechanisms (easy to implements, not too error prone, or not too out of sight of DSSSL) to dynamically change the script based on the new input document being processed. The actual constructs do not allow this. Here is the status of the work done and which integrates several comments made on this list recently. Any comments? Your comments may help the reflections about the subject and help create a better implementation. several minds sharing their thought is often better than a single one. Science evolved that way ;-) talking for the OpenJade team. regards Didier PH Martin mailto:martind@xxxxxxxxxxxxx http://www.netfolder.com DSSSList info and archive: http://www.mulberrytech.com/dsssl/dssslist
Current Thread |
---|
|
<- Previous | Index | Next -> |
---|---|---|
RE: About multiple output documents, Didier PH Martin | Thread | RE: About multiple output documents, Avi Kivity |
About multiple output documents fro, Didier PH Martin | Date | RE: One file into many, Didier PH Martin |
Month |