Felix S.

> This GCC extention generates for all architectures code which workaround problems with unaligned access. I didn't know that attribute had that effect. Does it also cover the case of...

Actually, it's not that hard to find: there's [an SO question](https://stackoverflow.com/a/8568441), which leads to [GCC PR #51628](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51628). And apparently direct access to packed fields is fine, and so is immediately...

`sector_size` is at offset 11 in the boot sector, `dir_entries` at offset 17 and `sectors` at offset 19. If the `packed` attribute weren't there, the compiler would assume a byte...

Use `\303`. Short file names are not stored as Unicode characters, but as bytestrings in the 'OEM code page', which is locale-dependent (long file names are stored as UCS-2 strings)....

https://sourceforge.net/projects/fatresize/ already exists.

With respect to `Aligned`: I remember people some time ago floating the idea of adding yet more pointer niches by recognising the fact that under some ABIs, certain pointer ranges...

Good point. But it also applies to modifying the ABI of regular references, which this proposal does without much restraint. In practice I wouldn’t expect `WellFormed` to be much different...

I don’t think this *should* be considered sound. If applied to ZSTs used as proof tokens, it would render them toothless, especially if they are `Copy`. And yes, I have...

They might as well be both 2 bytes: with variant `Z` having bit representation `0x0000`, `P1` being `0x0001`, `N1` being `0x0003`, `N2` being `0x0005`, `N3` being `0x0007`. There is no...

As for `dangling`: I thought the point of the niche optimisation is to remove the need for such a thing. But in the worst case, it could be simply declared...