[
https://issues.apache.org/jira/browse/XERCESC-1652?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12603876#action_12603876
]
Jesse Pelton commented on XERCESC-1652:
---------------------------------------
I'm not tracking 3.0, but I suspect so. I see that you've changed how statics
are initialized and cleaned up, but the problem I encountered was with method
calls on the static document, not initialization. Among the calls made using
the global object:
DOMDocumentImpl::cloneString()
DOMDocumentImpl::getPooledString()
new (DOMDocument *) DOMNamedNodeMapImpl(this)
new (DOMDocument *, DOMMemoryManager::DOCUMENT_TYPE_OBJECT)
DOMDocumentTypeImpl()
I haven't looked at how these are implemented lately. I don't know for sure
that they all use member data, but it seems safest to assume that they do. If
they don't now, they could at some future time.
> DOMDocumentTypeImpl is not thread safe
> --------------------------------------
>
> Key: XERCESC-1652
> URL: https://issues.apache.org/jira/browse/XERCESC-1652
> Project: Xerces-C++
> Issue Type: Bug
> Components: DOM
> Affects Versions: 2.7.0, 2.8.0
> Environment: Multithreaded programs
> Reporter: Jesse Pelton
> Attachments: DOMDocumentTypeImpl.diff
>
>
> DOMDocumentTypeImpl.cpp uses a static gDocTypeDocument() method to return a
> singleton doctype document that provides various utility services (allocating
> named node maps, cloning strings, and so on). Any calls to methods of this
> static document that can modify its instance data, including calls that
> allocate memory, must be synchronized to keep the document's state consistent.
> Patch against the trunk to follow, but note that it should be reviewed
> carefully. First, I pretty much blindly copied the synchronization code from
> DOMImplementationRegistry.cpp and may have gotten something wrong. Second,
> the patch is adapted from changes I made to v2.7.0 code and has not been
> tested (or even compiled) (Updates to the trunk prevent successful
> application of a patch against v2.7.0 to code from the trunk.) And lastly,
> it's possible that the trunk updates obsolete my patch, though it still seems
> relevant.
--
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]
