[ 
https://issues.apache.org/jira/browse/XERCESC-1821?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12615855#action_12615855
 ] 

David Bertoni commented on XERCESC-1821:
----------------------------------------

I agree with Scott that (a) and (b) are non-issues, as long as we document that 
setting the default value is not synchronized.

As for Scott's question regarding (c): There is no way now to change the 
default value.  Rather, after constructing a DOMDocumentImpl instance, you can 
call DOMDocumentImpl::setMemoryAllocationBlockSize().  That means you need to 
call it for every document you create, if you don't like the default block size.

I doubt that the performance improvement of initializing one member variable 
from a constant rather than a static variable is even measurable given what 
else goes on in the constructor.

> Configurable  kInitialHeapAllocSize in xerces
> ---------------------------------------------
>
>                 Key: XERCESC-1821
>                 URL: https://issues.apache.org/jira/browse/XERCESC-1821
>             Project: Xerces-C++
>          Issue Type: Improvement
>          Components: DOM
>         Environment: All OS
>            Reporter: george
>            Priority: Minor
>             Fix For: 3.0.0
>
>
> The default  DOM initial heap size is 16k in the latest release (earlier 
> release was 64k).  this number could be still too big or too small for some 
> system.
> The idea is to make the following parameters configurable in the constructor 
> via a singleton read function.
> static XMLSize_t kInitialHeapAllocSize =  0x4000;
> static XMLSize_t kMaxHeapAllocSize     =  0x20000;
> static XMLSize_t kMaxSubAllocationSize =  0x1000;  // Any request for more 
> bytes
> Now the question is if you think this is a valuable suggestion?   I have a 
> working code and would like to share if you think it worth the effort.
> Cheers
> George

-- 
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