## 0.75.0

### Features

* ✨ Add support for custom `generate_unique_id_function` and docs for generating clients. New docs: [Advanced - Generate Clients](https://fastapi.tiangolo.com/advanced/generate-clients/). PR [#4650](https://github.com/tiangolo/fastapi/pull/4650) by [@tiangolo](https://github.com/tiangolo).

## 0.74.1

### Features

r) defer drbgs.Put(drbg) for len(b) > 0 { size := min(len(b), maxRequestSize) if reseedRequired := drbg.Generate(b[:size], additionalInput); reseedRequired { // See SP 800-90A Rev. 1, Section 9.3.1, Steps 6-8, as explained in // Section 9.3.2: if Generate reports a reseed is required, the // additional input is passed to Reseed along with the entropy and // then nulled before the next Generate call. entropy.Depleted(func(seed *[48]byte) { drbg.Reseed(seed, additionalInput) }) additionalInput = nil...

drbgPool.Put(drbg) } }() for len(b) > 0 { size := min(len(b), maxRequestSize) if reseedRequired := drbg.Generate(b[:size], additionalInput); reseedRequired { // See SP 800-90A Rev. 1, Section 9.3.1, Steps 6-8, as explained in // Section 9.3.2: if Generate reports a reseed is required, the // additional input is passed to Reseed along with the entropy and // then nulled before the next Generate call. drbg.Reseed(getEntropy(), additionalInput) additionalInput = nil continue } b = b[size:] } } var testingReader...

*   RNG behavior change for `tf.keras.initializers`. Keras initializers will now
    use stateless random ops to generate random numbers.
    *   Both seeded and unseeded initializers will always generate the same
        values every time they are called (for a given variable shape). For
        unseeded initializers (`seed=None`), a random seed will be created and