[ 
https://issues.apache.org/jira/browse/XERCESC-1737?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12520319
 ] 

Boris Kolpackov commented on XERCESC-1737:
------------------------------------------

Scott,

I tested your patch on various workloads and there was about 10% performance 
hit on larger documents (~300kb). I, however, was able to eliminate this 
slowdown by doubling InitialHeapAllocSize and MaxHeapAllocSize. Will it still 
be acceptable to you (i.e., the smallest DOM document will use 16Kb of memory)?

> Patches for the 2.8.0 release
> -----------------------------
>
>                 Key: XERCESC-1737
>                 URL: https://issues.apache.org/jira/browse/XERCESC-1737
>             Project: Xerces-C++
>          Issue Type: Improvement
>            Reporter: Boris Kolpackov
>         Attachments: heap.patch, patches-2.7.tar.gz
>
>
> I have prepared a patch for the following issues:
> * Support for g++ on AIX and HP-UX.
> * Auto-detection of aCC6 vs aCC3 in runConfigure.
> * Verbose mode (gmake VERBOSE=1)
> * Changed optimization level for GNU/Linux from -O to -O2
> * Fixes for 64 bit and SO_NAME options.
> * Project files for VS 2005.
> * Support for 64 bit Windows build.
> * Optimized DOM parsing (25-30% improvement).
> It consists of the following parts:
> ChangeLog        - description of the changes.
> xerces-2.7.patch - patch against the current xerces-2.7 branch
> VC8              - MSVC8 project files, should be copied to Projects/Win32/
> I have tested this code with most of the platforms mentioned in the
> "Xerces-C++ 2.8.0 release plan" email I send the other day.
> I would appreciate it if one of the Xerces-C++ developers review and
> commit this patch.

-- 
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]

Reply via email to