#727 running docbook xsl never returns

output: all formats
closed
nobody
XSL (1066)
5
2006-02-09
2005-12-12
Nils Meier
No

Hi

I have a rather docbook document (1.7MB) that I can't
run through docbook xsl... the operation never finishes.

I know it has something to do with indexterms/index but
this simply seems to work for smaller files whereas the
attached fails.

Can anyone else please run the docbook through and
report whether he see's the same problem? At the moment
I can only assume an endless-loop in the xsl ... but
sadly I don't know how to debug that myself.

Thanks
Nils

Discussion

  • Nils Meier

    Nils Meier - 2005-12-12

    docbook (zipped) that I can't transform

     
  • Robert Stayton

    Robert Stayton - 2005-12-28

    Logged In: YES
    user_id=193218

    Did you try setting the stylesheet parameter index.on.type
    to 1?

    This is an unusual document because it has 11 indexes with
    a type attribute. If you did not set that parameter, then
    each index gets all indexterms. Index processing is very
    intensive, and doing 11 full indexes could indeed take a
    long time.

    To test, I commented out all <index> elements, and
    processed without setting the above parameter. Your
    document processed with xsltproc in 0:48 (48 seconds).
    When I added the first index, it took 2:33. Doing 2
    indexes took 5:09. It seemed to be scaling linearly per
    index at about 1:35 per index. How long did you let it
    run?

    When I set the parameter to 1, I could successfully
    process the document with 2 indexes in 2:23, and all
    indexes in 12:46.

    Please let me know if this parameter works for you.

     
  • Nils Meier

    Nils Meier - 2006-01-15

    Logged In: YES
    user_id=118458

    Bob,

    thanks for having a look at this. I've just downloaded
    'a clean' 1.69.1 and re-ran this test with saxon 6.5.5
    like so

    $ java com.icl.saxon.StyleSheet -o foo.fo foo.xml
    ./docbook-xsl-1.69.1/fo/docbook.xsl index.on.type=1

    without index.on.type I can see what you're describing
    about a minute of time per index.

    With index.on.type=1 processing the document takes
    3.5 hours. Interestingly the output foo.fo grows to
    9.44 MB within a couple of minutes while the end-
    result is just 9.69 MB ... the time is indeed spend
    on generating the indexes.

    I'll do some more testing but I wonder why this
    excessive amount of time is required - will report
    with more numbers.

    Could this have something to do with the version
    of saxon/Java?

    Regards
    Nils

     
  • Robert Stayton

    Robert Stayton - 2006-02-08

    Logged In: YES
    user_id=193218

    I'd like to close out this bug report. Is there any
    reason not to?

     
  • Nils Meier

    Nils Meier - 2006-02-09
    • status: open --> closed
     

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.





No, thanks