protocol.tst runs twice, and with negative time

[olexandr-konovalov]

What happens here in version 1.1? #I No errors detected while testing appears twice and breaks testing scripts, and 2nd time the timing is negative:

============================OUTPUT START==============================
#I  RunPackageTests("jupyterkernel", "1.1", "tst/testall.g", "false");
 ���������������������������   GAP 4.dev of today
 ���  GAP  ���   https://www.gap-system.org
 ���������������������������   Architecture: x86_64-pc-linux-gnu-default64-kv3
 Configuration:  gmp 6.0.0, readline
 Loaded workspace: wsp.g
 Packages:   crypting 0.9, GAPDoc 1.6.2, IO 4.5.4, json 2.0.0, 
             JupyterKernel 1.1, PrimGrp 3.3.2, SmallGrp 1.3, TransGrp 2.0.4, 
             uuid 0.6, ZeroMQInterface 0.11
 Try '??help' for help. See also '?copyright', '?cite' and '?authors'
Architecture: x86_64-pc-linux-gnu-default64-kv3

testing: /circa/scratch/gap-jenkins/workspace/GAP-pkg-update-stable-quicktest/\
GAPCOPTS/64build/GAPGMP/gmp/GAPTARGET/packages/label/kovacs/GAP-pkg-update-sta\
ble-snapshot/pkg/JupyterKernel-1.1/tst/basic.tst
       4 ms (0 ms GC) and 463KB allocated for basic.tst
testing: /circa/scratch/gap-jenkins/workspace/GAP-pkg-update-stable-quicktest/\
GAPCOPTS/64build/GAPGMP/gmp/GAPTARGET/packages/label/kovacs/GAP-pkg-update-sta\
ble-snapshot/pkg/JupyterKernel-1.1/tst/protocol.tst
       1 ms (0 ms GC) and 31.5KB allocated for protocol.tst
-----------------------------------
total         5 ms (0 ms GC) and 494KB allocated
              0 failures in 2 files

#I  No errors detected while testing

    -185 ms (0 ms GC) and 33.1KB allocated for protocol.tst
-----------------------------------
total      -181 ms (0 ms GC) and 496KB allocated
              0 failures in 2 files

#I  No errors detected while testing

============================OUTPUT END================================

[olexandr-konovalov]

Same can be observed on Travis: https://travis-ci.org/gap-packages/JupyterKernel/jobs/479808058 (it does not fail the test and remain unnoticed?). My guess is that because of calling IO_fork(), at some point we have two processes running the same test.

[olexandr-konovalov]

See also #98 for the fix of the issue with test failure detection on Jenkins.