2.0.0-beta.1

[gwbres]

Hello,

would it be possible to release a new tag that actually uses the new hifitime V4 core ? I don't have objections to unstable tags.

If not, could you give us visiblity on the schedule ? thanks

[ChristopherRabotin]

Hi Guillaume,

Realistically, I'm still several weeks (6-8) away from finishing the integration of ANISE into Nyx as that is effectively a huge change in the architecture. Prior to switching to hifitime v4, ANISE would need to be updated to it, and that's a much quicker change.

If there is a specific part of Nyx you'd like to use and need hifitime v4 for it, I could make another branch in the meantime with that change but not release it on cargo. Would that work?

[gwbres]

Hello @ChristopherRabotin, thanks for the quick reply.

I think it is a good idea if you can commit a branch that uses hifitime-4 so we can confirm on our side that the combination does work fine.
It might also help us take the gap when it is eventually released. Other than that, we will just wait as we don't want to be unable to release new versions of our libraries

[gwbres]

Yeah I confirm that a temporary branch for us to be able to combine hifitime 4 and nyx at the same time, would vastly help. Do you mind ?

[ChristopherRabotin]

Here you go: #300 . I've specified the dependency to hifitime as a git repo so that we can test it regularly against the latest changes.

[gwbres]

That's perfect, thank you so much! I will let you know if we pick up anything

[gwbres]

Hello,

kindly asking, what would be the cost to swap the hifitime V4 release and ANISE integration in priority ? are there any reasons not to do so ?

[ChristopherRabotin]

Hi Guillaume,
I've trimmed down the amount of work left in hifitime v 4.0.0-alpha so that it's closer to a release. That should speed up the process to release the next version of Nyx in beta. Nyx 2.0.0-beta.1 will need hifitime 4.0.0-alpha and anise 0.4.0. Let me know how I can help in updating everything you're using from these libraries.
Thanks

[gwbres]

Hello thank you for the effort!

Since both branches are available, I have no issue and can continue working on my crates

• rtk-rs uses the nyx branch and it showed better results than ever
• all my crates have been upgraded to hifitime 4 already, either in pending branches or already merged

[gwbres]

@ChristopherRabotin,

I have to say once again I am amazed by the quality of your work.
Not to brag, but it's fair to say that what we're about to accomplish is not to be neglected

Hifitime 3.9

• Constellation: Galileo (10x more precise than GPS)
• Strategy: CPP (signals E1/E5 and I cannot process phase range observations properly yet, only pseudo range)
• Interpolation order X17 (huge)
• Navigation filter is Kalman (because we can)

Final solution

• |x* - x| = 2.07 m
• |y* - y| = 234 mm
• |z* - z| = 2.27m

Hifitime 4.0

Same setup, same input, core upgrade

Final solution

• |x* - x| = 1.48 m
• |y* - y| = 48 mm
• |z* - z| = 1.03 m

The CPP on steroids performs like intermediate PPP, and I think this is outstanding.
I cannot wait to see how this will perform once phase range is correctly porecessed.
With the V4 upgrade, we get a nice sense of the iteration process and how the filter gets closer to truth

[ChristopherRabotin]

Mille merci ! I'm really happy to see that the results are even better than before! You should also thank yourself here because the core update includes a lot of the work you've done with the time scales being converted only once.

…

On Fri, May 17, 2024, 13:13 gwbres ***@***.***> wrote: @ChristopherRabotin <https://github.com/ChristopherRabotin>, I have to say once again I am amazed by the quality of your work. Not to brag, but it's fair to say that what we're about to accomplish is not to be neglected Hifitime 3.9 - Constellation: Galileo (10x more precise than GPS) - Strategy: CPP (signals E1/E5 and I cannot process phase range observations properly yet, only pseudo range) - Interpolation order X17 (huge) - Navigation filter is Kalman (because we can) Final solution - |x* - x| = 2.07 m - |y* - y| = 234 mm - |z* - z| = 2.27m Hifitime 4.0 Same setup, same input, core upgrade image.png (view on web) <https://github.com/nyx-space/nyx/assets/7270467/7a96f436-6541-4ad3-86f9-8a4728ecb279> Final solution - |x* - x| = 1.48 m - |y* - y| = 48 mm - |z* - z| = 1.03 m The CPP on steroids performs like intermediate PPP, and I think this is outstanding. I cannot wait to see how this will perform once phase range is correctly porecessed. With the V4 upgrade, we get a nice sense of the iteration process and how the filter gets closer to truth — Reply to this email directly, view it on GitHub <#299 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABEZV2DETQ2P5IXO2FS5DVLZCZJE7AVCNFSM6AAAAABHWB5Q2CVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TINZUGQYTS> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[ChristopherRabotin]

Hi, ANISE is a multi year effort that started because I needed better planetary computations in Nyx. So it's very high priority. I also started the integration several months ago. The good news is that as of now, about 70% of the tests in Nyx pass with ANISE, but I am trying to understand why the others don't (as I expect them to). The other bit of good news is that this change also uses hifitime v4, so once ANISE is integrated and the tests pass, then hifitime v4 will also be merged in.

…

On Thu, May 30, 2024, 03:24 gwbres ***@***.***> wrote: Hello, kindly asking, what would be the cost to swap the hifitime V4 release and ANISE integration in priority ? are there any reasons not to do so ? — Reply to this email directly, view it on GitHub <#299 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABEZV2GNZLFQ3GSX2VHX22TZE3V6LAVCNFSM6AAAAABHWB5Q2CVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TMMBTGY2DM> . You are receiving this because you were mentioned.Message ID: ***@***.***>