Fixes for align_offset

[pvdrz]

This addresses @RalfJung's comments in #945

[RalfJung]

Thanks!

@bors r+

[bors]

📌 Commit e2c54e6 has been approved by RalfJung

[bors]

⌛ Testing commit e2c54e6 with merge b625812...

[RalfJung]

Do we also have a more direct test that immediately calls align_offset? That seems useful to have.

[bors]

☀️ Test successful - checks-travis, status-appveyor
Approved by: RalfJung
Pushing b625812 to master...

[pvdrz]

we used to but then we disabled the intptrcast part and removed it Sent from ProtonMail mobile

…

-------- Original Message --------

On Sep 28, 2019, 13:38, Ralf Jung wrote: @RalfJung commented on this pull request. --------------------------------------------------------------- In [tests/run-pass/aligned_utf8_check.rs](#957 (comment)): > @@ -1,13 +1,6 @@ fn main() { const N: usize = 10; - - let x = vec![0x4141u16; N]; - - let mut y: Vec<u8> = unsafe { std::mem::transmute(x) }; - unsafe { y.set_len(2 * N) }; - - println!("{:?}", std::str::from_utf8(&y).unwrap()); - - let mut x: Vec<u16> = unsafe { std::mem::transmute(y) }; - unsafe { x.set_len(N) }; + let vec = vec![0x4141414141414141u64; N]; + let content = unsafe { std::slice::from_raw_parts(vec.as_ptr() as *const u8, 8 * N) }; + println!("{:?}", std::str::from_utf8(content).unwrap()); Do we also have a more direct test that immediately calls align_offset? That seems useful to have. — You are receiving this because you authored the thread. Reply to this email directly, [view it on GitHub](#957?email_source=notifications&email_token=AHSUMUAEBF36Q3EVYNVEMVTQL6QBVA5CNFSM4IXSUCWKYY3PNVWWK3TUL52HS4DFWFIHK3DMKJSXC5LFON2FEZLWNFSXPKTDN5WW2ZLOORPWSZGOCGH72QQ#pullrequestreview-294649154), or [mute the thread](https://github.com/notifications/unsubscribe-auth/AHSUMUG4JBCW2WAFW76RRBDQL6QBVANCNFSM4IXSUCWA).

[RalfJung]

I don't see how that relates?

We could have something that creates a Box<[u32; 4]> (should be 4-aligned on all platforms) and plays around with it a bit, like aligning some pointers in there to various alignments and see that they all behave as expected.

[pvdrz]

oh no it's not like it was mandatory or anything. I'll add it later :) Sent from ProtonMail mobile

…

-------- Original Message --------

On Sep 28, 2019, 14:19, Ralf Jung wrote: I don't see how that relates? We could have something that creates a Box<[u32; 4]> (should be 4-aligned on all platforms) and plays around with it a bit, like aligning some pointers in there to various alignments and see that they all behave as expected. — You are receiving this because you authored the thread. Reply to this email directly, [view it on GitHub](#957?email_source=notifications&email_token=AHSUMUF2PG2XN4EX2EDFI7DQL6U3HA5CNFSM4IXSUCWKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD73AWMQ#issuecomment-536218418), or [mute the thread](https://github.com/notifications/unsubscribe-auth/AHSUMUHBG5BF6VJAYQZMPVLQL6U3HANCNFSM4IXSUCWA).