Data corruption with read() / ReadFile()

[caduel]

You are aware that the 100ms timer is not guaranteed to "fire" within 100ms?
(Linux is not a real-time OS.)

[kuzulis]

And why use a timer to call the reading?
It is easier to do - is to use ready-made libraries for use with serial port:
QextSerialPort
or
QSerialDevice v 0.1.0 (but better from SVN) :
release: http://fireforge.net/frs/download.ph...ice_v0.1.0.zip
svn:http://fireforge.net/snapshots.php?group_id=199

[Teuniz]

And why use a timer to call the reading?
It is easier to do - is to use ready-made libraries for use with serial port:
QextSerialPort
or
QSerialDevice v 0.1.0 (but better from SVN) :
release: http://fireforge.net/frs/download.ph...ice_v0.1.0.zip
svn:http://fireforge.net/snapshots.php?group_id=199

I tried QextSerialPort one or two years ago but it wouldn't compile and it was not maintained anymore.

I will have a look again at these libs but I want to know why Qt does not work well with direct call's to read()...

[kuzulis]

Teuniz ,

used only QSerialDevice (from SVN).

Your questions difficult to answer, because you did not show the code.

If you have any questions QSerialDevice - you can write to me at mail (e-mail address in source code QSerialDevice)

[Teuniz]

You are aware that the 100ms timer is not guaranteed to "fire" within 100ms?
(Linux is not a real-time OS.)

Yes, and I checked how many bytes are received during every call to read().
The device sends 772 bytes per second to the pc. Baudrate of the serial port is 115K2.
What I see is that one call receives 78 bytes, the next call 0 bytes, next call 78 bytes, next call 0 bytes, and so on.

When I plug-in the USB-stick, the amount of bytes received per call does not change which gives me the impression that the timer does not delay too much.

What happens is that in one block of data (772 bytes) 4 bytes are missing.