[
https://issues.apache.org/jira/browse/XERCESC-1732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12518164
]
Scott Cantor commented on XERCESC-1732:
---------------------------------------
The original patch was mine. Xerces 3.0 was modified to permit application
control over the heap sizes, this was my workaround until that release is done.
Yes, it "only" benefits small documents, but that's the vast majority of many
XML workloads today, and more importantly it is *mandatory* for small
documents. The normal heap size is completely unacceptable for long running
processes that might wish to hold documents in memory.
I think you would find that the majority of people would prefer this heap size
over the original, but I don't know that it's advisable to be patching the
2.7.0 sources while creating a 2.7.0 package. When I did it, I rev'd my copy to
2.7.1.
> reducing allocation sizes in DOMDocumentImpl.cpp
> ------------------------------------------------
>
> Key: XERCESC-1732
> URL: https://issues.apache.org/jira/browse/XERCESC-1732
> Project: Xerces-C++
> Issue Type: Improvement
> Components: DOM
> Affects Versions: 2.7.0
> Environment: Debian gnu/Linux
> Reporter: Jay Berkenbilt
> Attachments: allocation.patch
>
>
> The debian maintainer of the Shibboleth packages forwarded a patch to me
> which I am attaching. This patch decreases the values of kHeapAllocSize and
> kMaxSubAllocationSize in DOMDocumentImpl.cpp. Evidently, decreasing the
> chunk size here has significantly improved performance of shibboleth. As far
> as I can tell, no one has reported this here. I see that this patch would
> not apply cleanly to the file on the trunk, and it appears that this stuff
> has already been tweaked for performance reasons, but perhaps someone can
> evaluate whether this is a sensible patch. The source of the path is someone
> I consider reliable, so I'm going to go ahead and patch the debian version of
> the packages. (I am the debian maintainer for the xerces packages.)
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
