Excellent thread ! I just now stumbled on.

It illustrates very well a lot of the cnc / motion control pitfalls/needs and real capacity and issues/desires from users vs support,capacity,ability. And cost.

Base:
-I have absolutely nothing against the chinese controllers, or USBcncxxx controllers, machmotion, acorn/centroid, linuxcnc/mesa etc..
The whole field is extremely fragmented..
and *all* solutions and suppliers have some/a lot of/major issues not working (well), and a major major lack of documentation and how-to docs and features-supported now.

-My personal interests are commercial-industrial .. using a csmio-ip-s + all possible extra add-ons for quite a lot of money.
And, also, simple add-ons with a great hw solution, in my case and opinion the +/- best lower cost solution polabs-pokeys stuff.

-I have dealt with china, extensively, often for a lot of money (containerloads of expensive stuff), about 27 years.
My experiences are generally very positive - but very different to other trade globally.
I import chinese ac brushless servos, 60V and 220V, in the EU, for example - they are excellent.

My opinion:
The chinese controllers are like (pre) alpha-versions of any modern sw/hw motion control solution.. with great compilers (avoids silly errors), and good hw (dirt cheap today).
Pretty good stuff, IF the sw and docs were adequate, and they are not.
Docs are much more important than features- as-is, now, for the controllers.

This also applies to more or less all other sw controllers for linuxcnc and machx stuff and proprietary stuff.

It is possible the chinese controllers are good enough for some use, like 3d printers, or basic routers, in volume.
If so, they will sell, continue to be available, improve.
Legacy, docs, backwards compatibility are highly doubtful.

? What happens if the controllers change a chip--controller ?

I mean, a separately written open-source firmware is the best possible thing.
Ever.
And hugely valuable. 100M$+++. ++10x+...

But it is extremely difficult.
And almost-certain to fail, unless commercial-managerial talent- investor money is put into it. Imo. Ime. Yes experience - having done stuff like this.
My point is positive, not negative.

If I had time, I would try to arrange the same - resources for the guys doing this firmware, more personnel, better planning, testing, hw, testing methodology, reference platforms, std tests, etc etc.
But the firmware / motion control task is*monumental* riddled with n dependencies..
which is why *every single competitive attempt* has failed so far..

and the 2 best near-term solutions in linuxcnc and mach4 are still struggling with endless stuff -- in details.
Both work very well for 99.x% of users, and good or better than industrial stuff, for 99% of use, with the right hw, for 99% of use.

But users, almost always, need the 1% of rare stuff sometimes, and suffer when it does not work *as documented*.

Inputs, outputs, excellent probing, file-io, logs, communication, multiple-toolposts, multi-spindle, spindle-synch, run-as-spindle /axis at need, s-curve acceleration, defined probing and homing, come to mind.
Fast css, easy/bugfree/docs screw mapping, for me and all high accuracy uses. Huge commercial market 1B$+ / year on this.
Ethernet, piped communication, easy db linkage.
And all with offsets, scaling, rotation, adjustable (5 ways) exact-stop vs set-feed, hopefully conditionals in g code.
Those are just the basic must-have important ones.

Multiple kinematics, mirroring spindles for easy gcode/cam on multi-spindles, multi-spindle support for simultaneous cutting on many axis at once,
graphics.
And all above multiplied exponentially by about 10 modal states in gcode.