Add OsRng::{seed, try_seed}

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)

[tarcieri]

Summary

Adds methods for seeding SeedableRngs as replacements for the trait methods SeedableRng::{from_os_rng, try_from_os_rng} which were removed in #1674.

Motivation

This is functionality that was otherwise lost migrating OsRng out of rand_core.

Details

Since OsRng is no longer in rand_core which defines the trait, we have to reverse the relationship and place the methods onto OsRng instead.

[newpavlov]

I think MyRng::try_from_rng(&mut OsRng) and MyRng::from_rng(&mut UnwrapOsRng) already work good enough for this.

[dhardy]

Yes; this doesn't add much.

An extension trait providing the original from_os_rng() method(s) would however:

• be near backwards compatible (requiring only an import)
• not require referencing both MyRng (whatever that is) and OsRng at the same time when seeding

For seeding a single RNG this is still not sufficient motivation, but for many uses it might be. (Or we could define such a trait in that benchmark. Still blocked by actually having access to OsRng there though.)