I am curious why things like this and the other 6 bugs I posted to do with custom projects did not crop up during basic testing. They are all mainstream bugs as far as I can tell and they surely would have arisen with the most basic functionality testing.. Can it be something local?
Attachment log_2019_01_17.log added
Attachment Error.jpg added
It feels like the more I use the application, the more it damages itself internally...sort of like when you get a serious disk error, and if you keep using the disk it rapidly propagates across the disk creating more errors...maybe its that ?! :)
Attachment Ranking.jpg added
Namely, I was building the strategies with a weighted fitness, and forgot to also put this weighted fitness ranking in each of the other project stages. I corrected this last night so each stage had the weighted fitness as a ranking. After that, things got really bad with the freezing.
I undid this one by one, starting with the crosscheck retest real tick, real spread, setting it back to return/drawdown ratio. And that fixed it. For every stage of the project, walk forward included. Even the builder stage runs significantly faster. All other stages though remain on the corrected weighted fitness ranking.
Going into the real tick, real spread settings, it now shows no selection at all for the fitness? See screen grab. But the project itself is now running smoothly.
This real tick, real spread retest crosscheck stage is one of the ones I have most issues with when it comes to the entanglement bug (https://roadmap.strategyquant.com/tasks/sq4_3972). For example, in addition to this bug, this stage absolutely refuses to retain the commission setting in data, no matter how I save the config, or project. It is as if this stage itself is corrupt, if such a thing is possible. I have no clue.
Now, it starts but doesn't really start, and cannot be stopped, but can be started, but cant because, apparently it is already started. But it inst. Nothing ever happens. Bizarre. See video.
Attachment project.cfx added
I 'cloned' the project, which appeared tohelp for a while...it ran for about 21 hours (8 cycles) before getting stuck again. Now its getting stuck randomly again.
Note, if you clone it, you need to reboot SQX before it will work at all. You cannot clone projectsthen run the new clone, you first need to reboot SQX. This applies to any custom project.
Anyway, the getting stuck bug is worse as it means custom projects cant be left to run on their own.
Attachment Stuck.jpg added
Attachment log_2019_01_22.log added
Attachment log_2019_01_21.log added
Attachment log_2019_01_20.log added
Attachment log_2019_01_19.log added
Gets stuck on 21 jan around 21:20
Status changed from New to Fixed
It is still not perfect, but we found the most possible reason - maxing memory usage in Java.
All data prepared at 08:09:51 and first WF test completes 3 seconds later at 08:09:51...this is normal, takes about 3 seconds per WF test, processors at pretty much 100% the whole time.
In the video you can see it gets stuck and never moves on. Processor activity drops to around 20% the whole time. It will remain in this state forever.
Sometimes it can complete a few more strategies thenget stuck. Sometimes a few project cycles then get stuck. But it will always get stuck somewhere in the project tree on some WF test at some point.
Rebooting computer and/or SQX doesn't solve anything.