nRF - fakes vs "genuine" incompatibility

I have recently received a nRF24L01+ board with PA and LNA. In my current project I replaced one of PCB antenna nRFs with this better and communication was lost. Luckily I recalled this comment from Hackaday. It claims there is an error in nRF Datasheet regarding NO_ACK bit in Enhanced Shockburst. Some clones follow the Datasheet while other work as the original nRF. Which sadly makes different nRFs incompatible. The linked comment says this incompatibility is triggered by enabling dynamic payload length by setting EN_DPL bit in FEATURE register. This does not break my nRFs. Only enabling W_TX_PAYLOAD_NOACK command by setting EN_DYN_ACK bit in FEATURE register causes the Enhanced Shockburst to fail. I don't know if the description in the linked comment is wrong or there is more different clones with different degrees of the "bug" but I though someone may be interested in this.

What you say is interesting and it would be nice to investigate.
As far as I stiffen these modules are pure pain.
I have about 10 of these modules of various versions and I have tested them on 4 different types of arduini (Uno (Elego and Chinese clone), Nano and Mega (Keystudio).
The behavior of these modules is different on each arduino.
The most absurd case I have when I have 2 modules connected on the Nano and on the One, if I use the library of TMRh20 the Nano freezes instantly. If I use the old ManiacBug library the nano does not freeze, but it does not receive any data (but it works on the Tx side).
I do not know what to think about these modules, but they have been a very bad experience and I would not recommend them to anyone.

Ah, I forgot, the only "working" configuration is when I connect two modules to the Arduino Uno Elego and the other to the dull Chinese clone (4 euros).