Streaming encoder: keep track of encoded bytes that weren't written.

[marshallpierce]

They'll be retried on subsequent writes, but this plays poorly with
write_all. See rust-lang/rust#56889.

Hat tip to #90.

[codecov-io]

Codecov Report

Merging #92 into master will decrease coverage by 0.33%.
The diff coverage is 88%.

@@            Coverage Diff             @@
##           master      #92      +/-   ##
==========================================
- Coverage   96.52%   96.19%   -0.34%     
==========================================
  Files          12       12              
  Lines        1412     1471      +59     
==========================================
+ Hits         1363     1415      +52     
- Misses         49       56       +7

Impacted Files	Coverage Δ
src/write/encoder.rs	88.78% <87.5%> (+0.68%)	⬆️
src/write/encoder_tests.rs	95.66% <88.63%> (-1.43%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 61c2005...d30c85e. Read the comment docs.

[marshallpierce]

@dignifiedquire any thoughts on this implementation? Does this address your issue?

[dignifiedquire]

My current code relies heavily on write_all, so I can't easily test it, but it does look like it achieves what it needs to.

[marshallpierce]

Would it be feasible in your codebase to have a write_all_safe defined for some WriteExtension trait that doesn't choke on Ok(0) and use that instead? I'm wondering if that's maybe something I could suggest in the docs for how to get past the write_all limitations while the rust core team figures out what a better path forward would be.

[dignifiedquire]

Yes I probably need to do sth like that. I'll see if I find some time to do this later this week

…

On 3. Jan 2019, 21:45 +0100, Marshall Pierce ***@***.***>, wrote: Would it be feasible in your codebase to have a write_all_safe defined for some WriteExtension trait that doesn't choke on Ok(0) and use that instead? I'm wondering if that's maybe something I could suggest in the docs for how to get past the write_all limitations while the rust core team figures out what a better path forward would be. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[nox]

There is something I'm confused about, why is it an issue for write_all to return WriteZero? Shouldn't callers just ignore that kind of error in that case?

[marshallpierce]

You mean, why does write_all behave the way it does? It doesn't make sense to me either.