QXmlStreamReader inefficient with large QByteArray content

When using QXmlStreamReader on a QByteArray directly, it assumes (correctly) that all bytes are available.

Unfortunately this has two implications:

• on scanning the XML header "<?xml ...>" (in order to detect the encoding) the complete QByteArray data is converted to UTF-16 the first time (using BOM detection or assuming UTF-8 as source encoding)
• after encoding detection ("<?xml ... encoding="ISO-8859-1" ...>", this happens a second time (now with the specified source encoding)

While this works fine for smaller data sets, this can be a problem with large in-memory XML data.

Workaround is to wrap the QByteArray in a readonly QBuffer. Then data is consumed in 8 KB chunks meaning much less heap usage and decoding effort for larger data sets:

QByteArray xml = getXml();

QBuffer buffer;
buffer.setData(xml);
buffer.open(QIODevice::ReadOnly);

QXmlStreamReader xmlReader(&buffer);

This could be mentioned in the documentation (e.g. "Performance and Memory Consumption"). Or maybe some buffer size threshold could be defined that would cause QXmlStreamReader to wrap the QByteArray automatically?