[
https://issues.apache.org/jira/browse/XERCESC-2206?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17133691#comment-17133691
]
Roger Leigh commented on XERCESC-2206:
--------------------------------------
There would certainly be no urgency in releasing what's on the master branch.
We could potentially stage the changes there and leave it until next year, and
maybe also queue up any breaking changes which couldn't be applied for
compatibility reasons before now for 3.2. This would permit us to do the work
without committing to support two releases at the same time, if that would be
acceptable?
In benchmarking my application code, I've found that over 50% of the total CPU
time could end up spent in transcoding, and a big part of that was conversion
of UTF-8 to UTF-16 as input to Xerces-C++ and then more for reconversion of the
output. If it were possible, I'd find much more value in UTF-8 end-to-end
without involving UTF-16 or UTF-32. But being able to use UTF-16 literals and
std::ustring directly would reduce the overheads by a fairly significant amount.
> Use char16_t and unicode literals to replace various XMLCh types
> ----------------------------------------------------------------
>
> Key: XERCESC-2206
> URL: https://issues.apache.org/jira/browse/XERCESC-2206
> Project: Xerces-C++
> Issue Type: Bug
> Components: Miscellaneous
> Affects Versions: 3.3.0
> Reporter: Roger Leigh
> Assignee: Roger Leigh
> Priority: Major
> Fix For: 3.3.0
>
>
> Currently, XMLCh can be a variety of 16-bit types depending upon the
> platform, from wchar_t, uint16_t, unsigned short, to char16_t.
> To reduce the platform-specific variability, fix XMLCh to char16_t, and also
> permit the use of u"" unicode string literals in the codebase. This will
> allow replacement of Unicode constants with direct use of literals.
> This will additionally reduce the size of the test matrix with only one
> character variant to test.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
