When subclassing QIODevice, you must call QIODevice::seek() at the start of your function to ensure integrity with QIODevice's built-in buffer.

http://doc.qt.io/qt-5/qiodevice.html#seek

For me, i'm not test it, it should be

if (!QIODevice::seek(pos)) {
        return false;
}
qint64 ioIndex = this->pos();
.
.
.

In D4422#82848, @anthonyfieroni wrote:

When subclassing QIODevice, you must call QIODevice::seek() at the start of your function to ensure integrity with QIODevice's built-in buffer.

http://doc.qt.io/qt-5/qiodevice.html#seek

For me, i'm not test it, it should be

if (!QIODevice::seek(pos)) {
        return false;
}
qint64 ioIndex = this->pos();
.
.
.

No

I understand code, it will work now, but to me it's a workaround. What about to call QIODevice::seek(0) in constructor, other unchanged to original code ?

In D4422#82875, @anthonyfieroni wrote:

I understand code, it will work now, but to me it's a workaround.

Why is it a workaround?

What about to call QIODevice::seek(0) in constructor, other unchanged to original code ?

Please, test stuff before suggesting it, saying "No" to every random thing you think may work is not fun.

In D4422#82977, @dfaure wrote:

So your patch adds a member variable which duplicates QIODevice::d->devicePos, but there's no other solution without adding more API to QIODevice.

I deem QIODevice API incomplete, and your patch correct given the circumstances.

Yeah, one can use the _p.h header that exposes devicePos but i thought depending on the private implementation wasn't cool.

Yeah using qiodevice_p.h is out of the question ;)

For info the commit log syntax for autoclosing is
Differential Revision: https://phabricator.kde.org/D4422