Reset stonewall timer to make it work again with running all phases

[JulianKunkel]

Second part of bugfix for #119 allowing to use all phases in one execution.

[glennklockwood]

Thanks for catching (and fixing) this. Is this something that we can add to the test suite to catch any similar regressions in correctness around this functionality?

[JulianKunkel]

I think what we do need is a performance regression monitoring. Once I will have hardware (few month), I will run them with the tests, that would have shown the abnormal high values hopefully. More testing of such kind is welcome! It adds to our old question: needing test-hardware & someone willing to contribute ;-( Am Mi., 19. Dez. 2018 um 23:09 Uhr schrieb Glenn K. Lockwood < notifications@github.com>:

…

Thanks for catching (and fixing) this. Is this something that we can add to the test suite to catch any similar regressions in correctness around this functionality? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#120 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AE1uyiG0L-vqQcZeLd1jzNNPy2hcOpHMks5u6scugaJpZM4ZbEcz> .

-- Dr. Julian Kunkel Lecturer, Department of Computer Science +44 (0) 118 378 8218 http://www.cs.reading.ac.uk/ https://hps.vi4io.org/ PGP Fingerprint: 1468 1A86 A908 D77E B40F 45D6 2B15 73A5 9D39 A28E

[johnbent]

Is the only way to test with performance regression? Unit tests or functional testing will not help with this?

…

On Wed, Dec 19, 2018 at 4:12 PM Julian Kunkel ***@***.***> wrote: I think what we do need is a performance regression monitoring. Once I will have hardware (few month), I will run them with the tests, that would have shown the abnormal high values hopefully. More testing of such kind is welcome! It adds to our old question: needing test-hardware & someone willing to contribute ;-( Am Mi., 19. Dez. 2018 um 23:09 Uhr schrieb Glenn K. Lockwood < ***@***.***>: > Thanks for catching (and fixing) this. Is this something that we can add > to the test suite to catch any similar regressions in correctness around > this functionality? > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub > <#120 (comment)>, or mute the > thread > < https://github.com/notifications/unsubscribe-auth/AE1uyiG0L-vqQcZeLd1jzNNPy2hcOpHMks5u6scugaJpZM4ZbEcz > > . > -- Dr. Julian Kunkel Lecturer, Department of Computer Science +44 (0) 118 378 8218 http://www.cs.reading.ac.uk/ https://hps.vi4io.org/ PGP Fingerprint: 1468 1A86 A908 D77E B40F 45D6 2B15 73A5 9D39 A28E — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#120 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AB89PoYaqcvoCKJcB_DeppBU0MlNVMYWks5u6sfWgaJpZM4ZbEcz> .

[glennklockwood]

A unit test could probably catch this in principle, but mdtest is not factored very well for unit testing. We also never decided on a unit test framework since the only proposed solution (#38) required pulling in C++.

Since this error was just a number coming out wrong at the end of mdtest, I guess it's unclear how a functional test would catch this too. Unit tests are the right answer, but it'll be a lot of effort to get that sorted out I think.