Fred's firmware development diary

[Fred]

Setting cells in RAM works :-)

But burning it down to flash didn't. Need to investigate that...

Disable datalog sending works too.

Fred.

[Fred]

Sean is working on the flash stuff. I've been working on memory stuff and testing. The latest code still works, and I've got good mantises for Ben/Aaron to work on that will help me make progress. not sure what I'm working on next, but probably some data/memory/serial stuff of some sort.

flash usage :

currently about 0x3000 or less from 0x8000 is being used in linear space. This could probably be whittled down further with a bit of thought. The paged stuff should be better arranged too to avoid swapping ppage where possible.

The following things are kept in linear space :

interrupt handlers
comms main rountine
table lookups/validates/set/get
flash writing stuff
main function

some of that can be cut down and moved. all other code is already paged out.

Good luck to Sean working on flash (very close, bug could be in my usage)
Good luck to Aaron working on serial parsing (should be easy to find/fix)

Fred.

[Fred]

Last night was the xmas christmas function for the company I work for. It was pretty fun most of the night, and everything was flash etc. I kinda stopped having fun when I noticed that half the place (gross exageration, but at least 4 guys, almost certainly more, and chics too) were snorting coke in the bathroom. Am I old and boring, or is that seriously uncool?

Either way, I didn't make the last train from London at 11:15pm and ended up sleeping on a mates kitchen floor in a sleeping bag after leaving at about 2:30am.

So, no work got done.

Pretty tired too, so may not get a lot done on the way home either.

We'll see.

Fred.

[Fred]

I was out on wednesday, I was out on thursday, and I've been out tonight (saturday). I did however get some good work done on the JSON stuff. I didn't actually write any JSON, rather I attempted to model what would be required, how many per other there would/could be and the links etc. I think I'm getting closer on that front.

Fred.

[Fred]

Today was a write off, hopefully tonight I can do some good work on something or other. I've been distracted by all sorts recently and need to regain some focus on the firmware and docs for it. Tomorrow I'll try to put in a good days coding at work as well as after work, perhaps I can get things release ready soon. I still need more testing though, and I need help from Ben and Aaron and Sean for that to be complete. I can always release untested I guess at this early stage and put in bug fixes before the next release. I might get an early one tonight as there was no stock of my usual caffeine hit and I'll be dead on the train tomorrow if I have a long night tonight.

Fred.

[Fred]

I've done some good work between the last post and now, but none will be getting done tonight, and none got done last night or this morning either.

http://www.diyefi.org/forum/viewtopic.php?f=13&t=535

Hopefully I can get some good stuff done tomorrow.

Fred.

[Fred]

13 hours sleep and I feel just fine :-) (IE, no week long hangover Dave ;-) )

Lots to do today, I need to do my income tax return VERY soon, and tidy up some paper work and also, of course, work on the hardware stuff and json spec.

Fred.

[Fred]

I did some work on the firmware CODE tonight! GASP...

Must have had good day coding and writing unit tests etc in Java with EasyMock to inspire me. EasyMock is very cool, but I'm dying to try out Mockito, that sounds even better!

The work I started on tonight was, as always, something I'd been putting off. Refactoring the banked variables into a pointer driven arrangement. I have to say, the math code is a LOT better. A LOT!

Not to mention it will make datalogging MUCH easier and overall be faster and take less space. You know you are onto a winner when you can say all of that ;-)

I'll get onto finishing that up tonight. Maybe I'll put out a half baked release again without hardware and data specs included as it's been too long. I could aim to knock out another release with appropriate docs before January and work solely on that from this release on. I kinda wanted to put out a joint release of functional tuner/freeems together, but due to my own delays that's looking unlikely for a little while longer.

Hats off to Sean, Jared, Aaron, Dave and Ben for all of their effort thus far this year!! You've all done a marvelous job of helping the community grow and prosper, as have many others too. Many thanks to everyone involved.

I note that 0.0.1 went out on February 18th, I think I'll throw a party about then. Hopefully there will be something practical to party over rather than just "one year". The trouble is of course unless some of you are feeling like flying here, it will either be a me only party, or a me and Dave party. I'm sure we could have a good time regardless :-)

Lots of work left to do, but it's growing quickly and we'll start to see some fruit soon enough :-)

Right, back to eclipse and then dinner!

Fred.

[Fred]

Refactoring complete, saved 500bytes of code size (and they thought they saved a lot by going to ASM LOL), runs faster, looks cleaner, etc.

Still need to properly test it, but it runs and behaves much as before, except shorter datalogs now.

May fire the raw ADC array into it now just for a laugh.

Back to the JSON / Hardware...

Fred.

[Fred]

Today I ...

Fixed pointer to a pointer bug introduced with banking changes
Removed extra copies of core and derived vars not required
Documented new pointer scheme in header file

Tested above changes and can confirm datalogging in binary is working mint, but that the tuner can't keep up :-)

Looking forward to the latest update from Aaron!

I also researched JSON schemas and learned how to do array defs in it. I'll be knocking out some more JSON tomorrow.

Fred.