Re:Driver OnStep (LX200 like) for INDI

James,

The internet problem is anoying and only a "no chance".

The French government started a plan to connect 100% of the territory with Optical Fiber within 2020, all payed by Goverment, European Union ... so by the citizen.
So far so good, the Fiber arrived in my village.
The problem starts when it comes to commercialization.
This Network is opened for any internet provider in two manners:
A - The internet provider uses the complete network infrastructure including active devices, costs a bit more but zéro investment
B- The internet provider uses only the Fiber, and he provides it's own active devices, more investment but they manage themselves the network (only use the physical Fiber)

This makes that all little internet providers already offer their services on the Fiber but all are oriented Multimedia and relative expensive
and the Big provider that have better offers (example Internt Only ,... )take a bit more time util they install their hardware.

I am in the second case where my provider will start only in september and I definitively want to stay with my provider for the following rteasons:
- Fixed IPV4 full stack (no shared ports)
- Fixed IPV6/64
- Reverse DNS
- VPN support
- Free Phone for most countries and free Mobile phone for main countries (My daughter lives in Hong Kong

And all wih a cost that is one of the cheapest ...
For all these reasons I prefer to stay patient

=====================
Concerning you push I must just tahnk you very much for all your efforts.
I will merge to my master but prefer to wait a bit before pushing to main Master.

Thanks again

For the sleew problem (Kstars/sychart remains in sleewing mode instead of return in tacking mode after weak goto (plate solving...), i have made a quick and dirty workaround:
I launch this script: github.com/plop3/WatchdogOnStep/blob/master/WatchdogOnStep.sh
It monitor the state of tracking and if mount is blocked in sleewing state (sleew but coordinates dont't change), the script send a track order to resynchronize Indi.

Does not work

plop3 wrote: For the sleew problem (Kstars/sychart remains in sleewing mode instead of return in tacking mode after weak goto (plate solving...), i have made a quick and dirty workaround:
I launch this script: github.com/plop3/WatchdogOnStep/blob/master/WatchdogOnStep.sh
It monitor the state of tracking and if mount is blocked in sleewing state (sleew but coordinates dont't change), the script send a track order to resynchronize Indi.

Does not work

The 'KStars remains in Slewing mode, and never goes back to tracking' problem has been solved in OnStep master for some time now. Basically there are a lot of code that went into the 'end of slew' logic that Howard added after earlier complaints from KStars users.

If you use the latest master, you should never see the problem.

I use Jasem's INDI packages, and his KStars Bleeding.

I just tested and it seems to work fine

There is one bug, but it's an Ekos one from what I can tell related to the mount model alignment tool, which is if there's a delay changing to slew, and it gets checked too quickly then Ekos/Kstars assumes it's done with a slew before the command even gets sent.

I've filed a bug for kstars. bugs.kde.org/show_bug.cgi?id=410094 as I believe that's where the issue lies.

And since I left this post up overnight, before hitting submit, it's already fixed (or at least looks like it) in kstars git, and if I do observing tonight, I'll see if I have any issues. Thanks Jasem!



One thing that should be good shortly, once I'm finished is the :Gu support, is that it should be a bit less talky over serial, which hopefully will reduce the time and maybe? the load on the microcontroller it takes for updates. (I've noticed sometimes that megas have a bit of a delay some times, which I presume is due to load on the microcontroller, it seems less of an issue on STM32)

Planned to be replaced by :Gu#:
:Gm# (Pier Side)
:GX90# (Pulse Guide rate, still would need to be called for any custom/non-standard rates (aka not 0.25/0.5/1x sidereal) which I'd have to play with)
:GX95# (AutoFlip)
:GX96# (Preferred Pier Side)
:$QZ?# (PEC Status)

Not Replaced (and currently called)
:%BD and :%BR (backlash, can probably be read only occasionally)
:GXE9 :GXEA (minutes past meridian)
:A? (Align Status)
:GX02 :GX03 (Can probably be reduced and only dealt with in certain situations)
:GR :GD (Not changing, Dec and RA coordinates, need to be called!)

Though I do plan on still calling :GU# for now, because it's a LOT easier to read in either the status tab or logs, even though there's everything in :GU should be in :Gu (Along with more)

One other thing is possibly to allow for higher bitrates, which may (or may not) help or hurt. Unfortunately, most all of that is in the serial connection, which gets pulled in via telescope, and as far as I can tell, doesn't allow for on-the fly, so it'd have to send the command to change :SBx# then disconnect and reconnect. (Most all LX200 derived mounts should have the ability to do that, as Meade has had it in the lx200 protocol from close to the beginning of the protocol (Or at least there's no notation of anything not supporting it, as there is with almost everything added later))

I'm still working on the multipoint thing, but that's still proving a bit annoying. (I've got a bug, or a misunderstanding of what's being done, as it's coming out fairly off compared to the mount calculations.)

Do note, it unfortunately isn't fixed, yet. I thought it was, but just had it happen again. >_<

james_lan wrote: There is one bug, but it's an Ekos one from what I can tell related to the mount model alignment tool, which is if there's a delay changing to slew, and it gets checked too quickly then Ekos/Kstars assumes it's done with a slew before the command even gets sent.

I've filed a bug for kstars. bugs.kde.org/show_bug.cgi?id=410094 as I believe that's where the issue lies.

And since I left this post up overnight, before hitting submit, it's already fixed (or at least looks like it) in kstars git, and if I do observing tonight, I'll see if I have any issues. <strong>Thanks Jasem!</strong>

This bug is still there. Very annoying ...

Mount Model will slew and solve a star, then go to the next, then solve and slew too fast. The image is a streak and not a point.

I am using the latest Ubuntu 18.04 packages from Jasem's repositories, built a couple of days ago.

It broke, but mostly because I'd been messing with enough stuff that wasn't in master. Then the stupid spaces/tabs thing. (Which is now all spaces, but Gah that was annoying!)

As azwing is having fun with his internet (/s) I just bumped the OnStep version and submitted version 1.7, with the following changes, but likely some I may have forgotten, or missed going through commits.
Version 1.7:
- Added support for Reporting Guide rate (to PHD2 among others)
- Updated Error codes to match up with Android/SHC (Unknown reserved for unknown, so Unspecified = Unknown on other platforms)
- Added descriptions to SlewRate to match, slider kept which matches OnStep values
- Support for up to 9 stars for alignment
- Changed align so the last step isn't the (Optional) Write to EEPROM
- Added support for polar adjustments, without having to redo the entire model. (:MP# command)
- Support for Full Compensation/Refraction only, and 1/2 Axis tracking
- Cleanups


I don't have the :Gu# settings done yet, or a few other things in progress, but I figured the above would be welcome by people. And gaffes about merging an old variable name aside, I think things have been fairly well tested.

I added a Qtimer for 5 seconds (likely should be adjustable) to my own local copy to test if that fixes it. If so I'll send it on up to kstars. (It's supposed to be clear tonight! HUZZAH!)

- Changed align so the last step isn't the (Optional) Write to EEPROM

Do you mean the text of the instructions change, or there is a functionality change somewhere? If the latter, please elaborate more.

- Added support for polar adjustments, without having to redo the entire model. (:MP# command)

If you have a polar alignment that has significant errors (say > 3'), and you do the RefinePA (:MP), then you are better off re-doing the polar alignment since your gotos will off if you don't.

- Support for Full Compensation/Refraction only, and 1/2 Axis tracking

So now we have a Single and Dual button next to each of Full and Refr?

One more thing: the units for the polar error in the Align tab are wrong. They are ' but should be ". I am sure I reported this to Alain way back when, and it was fixed. Perhaps overwritten by some other change.

Thanks for your continued work on this ...

kbahey wrote:

- Changed align so the last step isn't the (Optional) Write to EEPROM

Do you mean the text of the instructions change, or there is a functionality change somewhere? If the latter, please elaborate more.

Yes, text changed.


Write align got set to Write Align to NVRAM/Flash (Actually should be EEPROM/Flash, but whatever, storage that doesn't go away with power!)

- Added support for polar adjustments, without having to redo the entire model. (:MP# command)

If you have a polar alignment that has significant errors (say > 3'), and you do the RefinePA (:MP), then you are better off re-doing the polar alignment since your gotos will off if you don't.

- Support for Full Compensation/Refraction only, and 1/2 Axis tracking

So now we have a Single and Dual button next to each of Full and Refr?

No, I did it a little differently, along with status, which shows Single Axis, 2-axis or N/A (If off)

One more thing: the units for the polar error in the Align tab are wrong. They are ' but should be ". I am sure I reported this to Alain way back when, and it was fixed. Perhaps overwritten by some other change.

See probably the first screenshot, Alain corrected this a while back. I don't think it got pushed up to the main one though. It's there now.

Thanks for your continued work on this ...

No problem, Open source is awesome.

Now if I can just find what the hell is with that bug in the alignment routine. (I've somewhat narrowed it down bringing out things like wireshark, but whatever the issue is, it seems to defy my finding it. Though I've narrowed it down some as to what gets called, but that shouldn't as far as I can tell. Somehow kstars thinks that the EqNP.s is busy when as per wireshark it's not, so it goes into the case of IPS_BUSY (== Slewing) when it's actually another status, but I don't see HOW. At least I know it's not the OnStep INDI driver side (or the INDI side, or the OnStep side... I think anyway ). )




Also: If anything breaks that was working before, PLEASE LET ME KNOW. There were a bunch of stupid spacing things that git decided a whole function constituted a change, some of which had actual changes in them, and I think I decoupled the spacing changes from actual changes, but... I'm only human, and a little bit more persistent than git.