Yes this is probably possible and already in the long term plan for pyXede. Shiv and I have already discussed a future where you can leave the datalogger running for a couple of days, then feed the logs to pyXede. You could then get a suggestions on how to adjust your map for better performance.

The import concept from your idea is that it'll make small relatively small changes over the virtual contiuum of tuning states. You could probably abstract the whole thing to some kind of linear optimization problem with the sensor readings as input. Then all you would have to is feed the datalogs to any number of free solver programs and say "solve it".

That being said, pyXede doesn't support map downloading/uploading/editing yet. I've asked Chiptorque repeatedly for the protocol for the map stuff. They are a little guarded about it because of their password function. They've expressed concern that if I had access to the full protocol I'd be able to break bypass the passwords easily. I don't know, as I've not seen it. Anyway people probably realize, this project advances in little spurts. I had a nice little productive period in july. Now I'm going back to school and got recruited for a big new project*.

donour

* For those you you interesed, I got hoodwinked into porting IBM's K42 operating system to 64bit X86 systems (i.e. opteron + EMT64). That's the whole thing; bring up, bootloading, memory management. Luckily the codebase supports linux device drivers, but still...

 

When hell is full, the dead threads will walk the earth....

Anyway. For those of you interested, I just posted a maintenance release to the sourceforge download site; aptly named v1.51 .This should fix a couple of problems folks were having with high speed logging. The SF site seems to be down at the moment, but keeping trying it.

The hardcore data analysis stuff still isn't in the codebase yet as it has been built as a seperate application, but it has neat features like "find noise ceiling". I haven't reverse engineered the mapping protocol yet either. The Xede is such a moving target :-p

d

 

so I tried the latest update.

It crashes python...

I can connect to the Xede, but when I attempt to generate a datalog for anything, dyno run or just to see if it works, I get an IO_ARRAY_ERROR or something like that.

I can generate some error logs if you want.

 

Yeah, email an error log to me.

By the way, which version of the xede firmware are you using?

d

 

I am using the very latest, it's only been in the car for a couple of weeks now.

I will try to get the logs to you this week. I have a very busy schedule which I am already late on...

I have not tried to do any debugging yet, didn't have the time. I will do some more when I get you the error logs.

 

so I got a chance to debug my problem. I got the software to work, but the graphing module is unreliable.

If I launch a road dyno session, it will crash the app, forcing me to kill the python shell.

I can run the dyno analysis tool, but it won't always work and often crashes the python environment. I don't think it's anything within your modules.

wx.gui - v1.13
pyxede - 1.11
dyno.py - 1.10
gnuplot..py - 1.1.1.1
rpmplot.py- 1.5

python 2.4.1

 

so are there any new developements with pyxede?

is pyxede ever going to be fully integrated into xede software?

 

I did some work to get it running on the new Intel Macs, but I doubt that very many people care about that.

Highly unlikely. I don't think that chiptorque is very interested. Plus, the whole point of pyxede is that it runs on Macs and Linux.

d