The best suggestion is to avoid using result tree fragments when you can.
I'm still working on a fix for this, which I hope to have available fairly
soon. The problem is not that the RTF is not released, it's that the
document factory that creates nodes for result tree fragments is not
releasing any memory.
One solution is to have the document release all memory when there are no
more RTFs in existence. The other, and more general solution, is for every
RTF to have its own document factory. This is more expensive for most
cases, but will always keep memory usage down.
Dave
|---------+--------------------------->
| | Holger Floerke |
| | <floerke@doctron|
| | ic.de> |
| | |
| | 04/14/2002 11:48|
| | PM |
| | Please respond |
| | to xalan-dev |
| | |
|---------+--------------------------->
>---------------------------------------------------------------------------------------------------------------------------|
|
|
| To: [EMAIL PROTECTED]
|
| cc: (bcc: David N Bertoni/Cambridge/IBM)
|
| Subject: Problem with growing memory (Xalan-C)
|
>---------------------------------------------------------------------------------------------------------------------------|
Hi there,
I have a serious problem with bug 7556. This Bug (or maybe this
enhancement potential :^) is responsible for the memory growing up during
transformation. David says there are ResultTreeFragments that can not be
freed so easily. How can I work around it?
a) Do *not* use call-template?
This is difficult. I thought about using a stylesheet to unfold my
"call-template", but there are recursions and maybe variables within the
template. This causes I can not unfold all "call-template" in the case of
recursion and have a problem related to names if I call a template more
than once in a template.
b) Explicit release a rtf?
Is there a possibility to write an extension function releases the
(known to be useless) rtf?
c) Use another XSLT-Processor?
Does anybody knows a good C based XSL(T)-Prozessor except Xalan-C and
xsltproc?
TIA
HolgeR
--
holger floerke d o c t r o n i c
email [EMAIL PROTECTED] information publishing + retrieval
phone +49 2222 9292 90 http://www.doctronic.de
