oversight.

conversions were mistakenly converting to/from milliseconds instead of
microseconds and thus wer...

impls will be properly grouped in the next commit, split from this one to
help keep a clean diff.

(split into two to keep diff cleaner)

oversight from v2.0 (42e7634bdcb470b274d4a9ee6b941a91f568364d).

`facility_to_mask()` was moved ...

perhaps i should deprecate it to discourage mistaken use?

overlooked in 10d17f8e699339857daeecb7660251f9e696feb9 and thus broken
since v1.2.

also fixes f...

been broken since v1.2: f6b1e22dbb63dc6f5c389480f0d0bdf7cbcae3e5

as explained in fa855e5f4a000a3a2e9d5ae4895e9144d6f97a84 it's pointless
now. supposedly crates.i...

some other trivial clarifications snuck in here also

at least one crate out there is using it; this helps users figure out
how it fits into the versi...

switch to having full stops for short entries

rendering on github plaes this text as a part of the preceedng bullet point.

the text added to the sys crates in c47b4fc2636036c51665bff85e0d34c87a95526d
was the copy from t...

since the conversion is fallible.

note, i tried leaving the `From` implementation in place with...

which forces callers to account for the possibility of later addition of
new variants in the way...

`error::PAErr` has it implemented, why not `error::Code` also

let the macro autogenerate suitable logic for correctly mapping the int to
an enum variant, maki...

such that they do not silently map to non-existent enum variants,
causing undefined behaviour du...

thus providing functionality for translating between enum variants and
integers.

this can then ...

the logic is somewhat non-trivial and should be tested

- assert that conversions work properly
- assert that conversions give the correct value mappi...

to support the next commit expanding testing enum variant mappings

objects are duplicated also because we want to attach methods to
them, not just to attach some/d...

missed in 1efaf361d8751d3eb057c2735ce80f2d50a5a175

as done with such badges on crates by the rust core developers.

otherwise, the badge, on gitbuh...

already stated elsewhere and removing saves some space / reduces clutter.

crates.io used to show some automatically per badge configuration in the
crate toml files, howev...

to prep for addition of badges to be displayed on crates.io pages (and in
github when user is vi...

left over from 59d1baa6c4282d6fcb4414ae2ace9c8e4463e93e

by enabling use of non-camel-case variants for the enum.

the underlying C API takes an i32. the param here in the binding was
`<Option<i32>` partly due t...

the default travis environment is Ubuntu Xenial which only has PA version 8.
Ubuntu Bionic has P...

some cases skipped over, particularly in binding, where it was deemed to make
reability worse or...

note, the devide type enum stuff was added without a feature guard, since
there's no value in hi...

(which introduces API changes that require a new feature guard)

(note, version.h just updated manually, may not be perfect)

these were never tested previously because when the code was written
`cargo doc` did not yet hav...

overlooked in bf9140aaa8e414b4798338515140d28cbd2201ba

'p' makes more sense for 'position' than 't' (where did 't' even come from?
the original C inter...

will help with simplification of link in doc generation under rustc 1.48+.

- lowercase 's' in 'shortcut'
- missing 'PA_' prefix

as per 39ff1c581e9fd47e0174f947673a0208818ad78c

`c_devs` was being comsumed by the for loop, making `c_dev_ptrs` invalid.

`eprintln!` does not appear in Cargo output; however, we can give Cargo
the warning instruction ...

(backwards compatible break, hence why not already done)

This reverts commit 6139dd3c3e286a5fd2...

the closures requires a param