On 8/7/19 4:29 PM, Andrew Dinn wrote:
On 07/08/2019 13:04, Dmitry Chuyko wrote:On 8/7/19 2:02 PM, Dmitry Chuyko wrote:Andrew, New code is buildable and new MapFail test passes on different platforms except it fails on linux i386:Ah, that even was x86_64 (sorry, mixed up results from automated system). I'll try to reproduce it by hand to see if there are any additional details.Whew! Thank goodness it is x86_64. You had me very worried when you said this was i386 :-) So, this is indeed a problem with the test. The internal detail message for the IOException will vary according to the Linux kernel release. If MAP_SYNC is not supported by the Linux kernel it will embed the message you saw. If it is supported the message will be the one the test checks for. I think the best thing is for the test not to bother checking for a specific message. The important thing is that we should get either IOException or UnsupportedOperationException. Checking for the message is not really necessary. I will remove this check from the test.
Ok, it will reduce complexity and likely let the test pass while still checking the same thing.
-Dmitry
regards,
Andrew Dinn
-----------
Senior Principal Software Engineer
Red Hat UK Ltd
Registered in England and Wales under Company Registration No. 03798903
Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander
