Investigate why ProB2-UI needs a higher TIME_OUT to work
Description
Attachments
2
relates to
Activity
Show:
dgelessus June 15, 2020 at 2:00 PM
This is probably fixed now as a side effect of - the ProB 2 UI now also reuses the same probcli instance for multiple machines, so the timeout should no longer occur here either when the machine is loaded for a second time.
Michael Leuschel April 3, 2019 at 11:21 AM
I think I found one explanation: ProB2-UI starts a fresh probcli everytime. The first time CLP(FD) is used there is some penalty due to the SICStus JIT running (that is my understanding). Hence, ProB2-UI runs into the time-out, ProB Tcl/Tk only runs into the time-out the first time; then the JIT/SICStus engine is "warmed up".
Fixed
Details
Details
Assignee
Unassigned
UnassignedReporter
Michael Leuschel
Michael LeuschelLabels
Priority
MajorCreated April 3, 2019 at 10:29 AM
Updated June 15, 2020 at 2:00 PM
Resolved June 15, 2020 at 2:00 PM
When using the example blocksworld_medium.mch from prob_examples
and adding the line
SET_PREF_TIME_OUT == 70; SET_PREF_STATIC_ORDERING == TRUE;
ProB Tcl/Tk can load the machine without a time out appearing, but in ProB2 UI we systematically get a time-out.
I have seen this with other models (AMASS) and the SETUP_CONSTANTS transition.