publishing an intermediate release 3.1.0?

Been playing for 1 hour. No problems, no crashes, no hang-ups. Tried "English" and "Nederlands". Tried the "Crash test" language file as well. No issues. Devo12S.

Just an idea:

Could it be that "Kreidler" has some remainder in memory from an earlier version that screws up his file system?

I noticed the Dfu tool does an erase loop first and then writes the new firmware and same for LIB. But when I change from original FW to Devention, most of the time my filesystem is still there and the same goes for the original FW. All model info is there. So that means this part of the memory has not been erased or cleared.

So what I am trying to say is that it may be necessary to do the same for that part of the memory that is not being used for FW and LIB. Do some kind of an erase/clean operation to make sure there are no un-expected remainders of previous downloads.

Of course this should be a user defined function, so you could choose to erase/clean this memory (or the whole memory) or not.

So you have the option to start with a "virgin" TX.

richardclli wrote: I spot that there is a range check feature in this version. Where can I find the entry point of this function?

A range check supported as a tx function would need to evaluate the RSSI signal from the receiver which is sent back to the transmitter.
As far as I know deviation never had such a feature.
Range checks are usually done by foot...

Pattaya01 wrote: Just an idea:

Could it be that "Kreidler" has some remainder in memory from an earlier version that screws up his file system?

When I read yesterday in the evening (local time...) that kreidler still has reboots of his Devo10 I was kinda shocked.
I have been going through the code for several times and I am as sure as I could ever be that no kind of weird language file can still lead to errors inside deviation - non existent, too short or too long, bizarre characters, control characters - nothing.
I began thinking that something is wrong with his tx. But he can force the resets simply by changing the language file - a tx failure should occur independently from the content of translations.

@kreidler: please don't feel offended with this question, but can you make sure that you have installed the actual version (should end with "beta-10-e11861f")? I am a little bit astonished that I have not received any confirmations for the crashes.
I know you have spent much time on testing so I am sure you know what you do but I simply don't have any other ideas now.

lasakro wrote: Using the German file that came with Beta 10 I spend about 15 minutes blindly going through the menus on the Devo 10 changing anything I could, saving and had no resets.

lasakro, the German file works due to shortened strings. Please use the 'English_Crashtest' file also coming with the beta10 filesystem and do the test as I described above.

rbe2012 wrote:

Pattaya01 wrote: Just an idea:
Could it be that "Kreidler" has some remainder in memory from an earlier version that screws up his file system?

...I began thinking that something is wrong with his tx. But he can force the resets simply by changing the language file - a tx failure should occur independently from the content of translations.
@kreidler: please don't feel offended with this question, but can you make sure that you have installed the actual version (should end with "beta-10-e11861f")? I am a little bit astonished that I have not received any confirmations for the crashes.
I know you have spent much time on testing so I am sure you know what you do but I simply don't have any other ideas now.

No problem at all. I like the testing but sometimes the development is so fast that at least I do not have the time to follow the updates in any case. This is hobby and my hobby is normally using this Tx and you know I asked you already to stop looking for this issue but I understand also why you continued.
For testing I normally do not compile by myself. In order to get same results from anybody I always use the files from the download section. Sorry, but no Tx on hand at the moment - so I assume to have df29c66 or 62f92b3 installed. Additionally to get a 'virgin' TX I always delete the filesystem and install the co-delivered one. After copying, of course, I put some modelx.ini and the tx.ini back again. But I can remember that PB wrote somewhere that the filesystem will not be affected when switching between origin firmware and deviation because of using different section of the memory.

rbe, we have lasakro with a devo 10. Hopefully he? will get through the menus with the special file later on.
In most cases (I will check later on again) this issue happened when the string for the menu lines was too long to be displayed on the LCD. On the sub config pages I was not able to simulate (e.g. with a long button caption) the crash. Just one idea, but I do not know if I understood the code correctly: One of the LCD files forces too long strings to be displayed in the next line. Did I understand correctly? Then it culd be an option to cut there.

English_Crash also crashed on my Devo 10, Beta 10.

lasakro wrote: English_Crash also crashed on my Devo 10, Beta 10.

Same results for me on Devo 10 but with all the latest commits from RBE (self compiled) post beta 10. I can go back and test with the actual beta 10 if you'd like.

To be specific, I changed language to the crashtest then scrolled down, once the first long line scrolled off the screen it crashed back to the splash screen and since I never exited the TX config I was back to the default English.

WheresWaldo wrote: To be specific, I changed language to the crashtest then scrolled down, once the first long line scrolled off the screen it crashed back to the splash screen and since I never exited the TX config I was back to the default English.

If you do not switch off the TX after changing the language you do not need to edit the TX.ini because the language change has not been saved to the file.

I think going back to the beta10 is not necessary for you because getting same results.

Just posted some additional tests in the ticket #2 area of rbe's fork.

WheresWaldo wrote: I can go back and test with the actual beta 10 if you'd like.

This will not be necessary - I didn't make any relevant changes after beta-10.

Thanks to all for testing - I will close the door and start again thinking and reading...

Well, it seems to be a specific Devo10 problem. The problem only occurs:

- Devo10
- Too long strings

So it could be related to some of the hardware on the Devo10 when you send a long string to the display. It might choke up the buffer from the display, which in its turn sets a flag or generates an interrupt or whatever which makes the FW reset.

The remedy is quite simple. Don't send long strings and all will be OK. Restrict the model name to XX characters to prevent that to cause a problem and every Devo transmitter should work fine.

I am afraid that, unless you have a Devo10 with a logic analyzer attached to almost every signal, you will have a very hard time to find this issue.

Pattaya01 wrote: Well, it seems to be a specific Devo10 problem.
...
I am afraid that, unless you have a Devo10 with a logic analyzer attached to almost every signal, you will have a very hard time to find this issue.

Could also happen on Devo7e if somebody does changes to the source because 7e does not support these language files.
I used the build-in analyser of my mind . As far as I can see I have been sucessful. Updates later on from rbe's side.

Yes, still building the zips...
I have analyzed the LCD routines before and made some changes, but kreidler has found that still unexpected values can occur. So I changed the checks.
This could have happened in Devo10 and Devo7e, not in the color screens because the way of writing to the LCD is completely different (depending on the LCD controller used).

kreidler, you will remember that I mentioned a possible connection with the sound routines. I analyzed the map file this morning and the sound parameters are lying directly behind the LCD buffer in RAM. Since the music output is interrupt driven changes to this can push the processor in an endless loop where it is caught by the watchdog who resets the tx.
All issues seem to be explainable. Thank you for finding it out.

There is a new version beta-11 in my repo ready for download (and the source is committed).
kreidler, I am sure you will test it even without my appeal, but you know best what to do to try crashing the Devo10 again. Your statement will be very important for me.

beta-11:
Ok from my side. No crashes anymore. Sorry for that.

YES!!!

Having been away from flying for some time now, may I request and, have a link for the newest files please?

ave1 wrote: Having been away from flying for some time now, may I request and, have a link for the newest files please?

Always look here:

bitbucket.org/rbe2012/deviation-3.1/downloads

Thank You!

rbe2012, I am just starting the process of collecting screen shots for all the GUI screens. I am thinking about just using the 8, 10 emulators since the 7/10, 8/12 are basically the same. If the 12 are has been reworked to use the additional screen space I will collect those also. Is that correct? To simplify the 7/10 collection I will only take screen shots of fully changed screens and not as each item scrolls up the screen.

Goes without saying, screen shots will be of the English menus only.

English as base language for the screen shots will be ok - everybody should be able to identify the labels mentioned in the text.
Devo7e and 10 have the same screen - the only thing is that the Devo7e does not have the ability to change the language which results in a slight different tx config screen which should be shown. I seem to remember that something was different with the new gui for Devo7e but I have to look again.
Devo 6 and 8 have the same screen and differ only in some switches (as 7e and 10 do) which also lead to different pages. But I personally have the impression that the Devo6 is quite rare in the filed so I would propose not to display the differences - better describing them once somewhere at the start.
Devo12 has some completely different screens (e.g. all on one page for the tx and the timer config where Devo8 uses scroll bars) - these should be shown.
If you want me to I can help you making the screenshots - just tell me which do you need.
Maybe we can screw screenshots from a scrolled page together so there is only one picture showing the full content.

Did you clone PBs or my documentation repo? I propose that you put the screenshots there (e.g. zipped in the downloads). I can clone that and send pull requests.