After a lot of work (for me and antigravity) : we have fixed a lot of things ... not very visual for everybody, but the core got a lot work (fully tested, admin part, emailing, secure, seo, ...). Yes, registering works again ! (it was broken since gae/google changes ;-( )

The future is brighter

So, here’s a simulator I threw together—using Gemini Canvas—to simulate a DJI Neo 2 with an RC3 controller (connected via USB) on a desktop. I think the result turned out pretty well; you can really feel the lack of power when trying to lift off the ground... It’s amazing what you can pull off in just 15 minutes.

As it stands, collision detection is limited to the ground level... but the RC3 integration, the rates, and the simulation itself are actually pretty solid.

dji neo2's simufpv (only playable with a real DJI RC3)

But I'm working on a more advanced one : simufpv as an htag2 app, which is more complete.

currently:

• playable with any real RC (dji or not), plugged with usb (on a desktop browser).
• playable on mobile, with touch support.

It will evolve soon ... but I need to decide if it goes thru htag2 or with the current(updated) pye... But there are so many things that are currently available (or broken) here. I will need to check all ;-( ... or restart from scratch.

Difficult choice ... But with IA, everything is possible ;-)

EDIT 26/09/11 : Finally pye is not dead. I (my llm) reach to adapt it, to be able to run htag2 app in live (here is an htag app (tetris) ... just copied/pasted from my main htag server). All others things should work as previously (but yes, there are a lot of old things whose are borken). It opens a lot of perspective (But no more workers and redys)

EDIT 2026/09/18 : Finally pye is dead ;-) The core of this website is based on "htserver" (a kind of fresh pye with great htag2 support). A page is turned ;-). I've reach to adapt old things on htserver (and evolve it). But sure, a lot of things should be broken (but might not be used since ages). I'll need to check all.

It's not completly finished, there are some bugs ... but prefer to release soon, it's available on render.com, in a free-plan (so may not be available when under quotas), but it's here:

https://atomic-bombers.manatlan.com/

And I’ve done all of this without writing a single line of code myself (to be honest, maybe 20 lines max out of the thousands upon thousands produced).

Everything is driven by antigravity and/or gemini-cli. Nowadays, my job clearly consists of 'conveying my intentions and guidelines... and carefully reading everything it tells me or generates.' I’m becoming something of a Product Manager or Solutions Architect.

The responsibility is on me to ensure it truly understands what I want—and I love it. It allows me to go much further without spending hours trying to grasp complex concepts I haven't mastered yet. I’m learning an enormous amount while saving a massive amount of time.

I’m not saying it’s pure magic—you still need to know how to express your intentions clearly and activate the right skill.md files—but the productivity gain is staggering.

Htagui, htbulma & htagweb needs htag 1.0.0 max !

and because, it's not christmas ... htag 2.0.0 is available too !

It takes years to see a "1.0.0" release, and now : 1.0.0 & 2.0.0 are here, in pypi ... crazy !

htag 1.0.0 is in end of life ... the future is htag 2.x.x

I don't know really where to go ... it's not really compatible with htag1, because some concepts are better implemented (kiss & rock-solid). But got a tone of ia/works, to be able to reach the feature of the first version (apk, etc ..). But the way is really really easier ...

I can't really mesure the gain of ai driven dev ... but it takes at least 1000x less times, to publish an htag2 (more solid) than the htag1. Ans as I said : a lot of things to do. But one thing is sure, htag2 couldn't exists if htag1 was not, because I've learn a lot of things with htag1 ... And with htag2, I continue to learn, by reading ia answer .... but a lot faster and easier !

Currently, I don't know how things will go ... but htag2 could easily be the way to follow.

The 1.0 will be the new major release soon. I've made some big "docs update" in htad docs, and htagapk ...

All old runners (BrowserHTTP, BrowserStarletteWS, ...) are deprecated, and will be removed in 1.0. There is now only one, which can emulate all features, with no dependancies, only pure python (no more starlette, uvicorn, tornado for specialized ones)

Now, I'll need to polish the api, and make the docs ...

Seems to be the best way, because google changes a lot of things, and buildozer follow the changes frequently. So i decided to use the right way of building an apk, using the latest changes from buildozer git.

The recipe is still there ... but the manual way is a lot simpler, using my new command line make.py ! It should work on any linux box (no windows!) But it's a lot simpler/faster !

And it should be more solid than the github action, which is currently broken ;-(

Hope it will live long !

Currently, I'm becoming a mojo addict ... And updated my sudoku tests with mojo/rust/c versions, to compare them. And I've created my first mojo issue ;-) I'm really thinking that can be the future of python programming.

htag is nearly ok for its 1.0 version. Its new web runner : htagweb works as well, but need polish ! I use it already, but want to migrate this host to full htag with htagweb soon ...

I'm developping a lot of htag's apps (the latest, is a gui for python/profiler named ppviz available on pypi) ...

I'd like to make a jbrout htag version, but I definitly have no times for that. And I believe than the best solution should be to make something like "beets" (for audio), but for photos. And startup a htag frontend. But no times, and I appreciate the "google photos" (while i'm interested in immich solution)

And since some months, I'm became a Nock addict too (vr/sport). It takes me a lot of time, but it's very very addictive ;-). You can find me here, every day ;-)