Saving files with sessionToken

[dplewis]

Closes: #1049

A small optimization with saving files in saveAll.
Clean up deferred promises.

@stevestencil Can you check and see if this works for you?

Let me know if anything else is needed.

[codecov]

Codecov Report

Merging #1057 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #1057      +/-   ##
==========================================
- Coverage   92.19%   92.18%   -0.01%     
==========================================
  Files          54       54              
  Lines        5150     5146       -4     
  Branches     1145     1148       +3     
==========================================
- Hits         4748     4744       -4     
  Misses        402      402

Impacted Files	Coverage Δ
src/promiseUtils.js	91.48% <ø> (ø)	⬆️
src/RESTController.js	84% <100%> (-0.22%)	⬇️
src/ParseObject.js	89.38% <100%> (-0.06%)	⬇️
src/TaskQueue.js	87.5% <100%> (-1.79%)	⬇️
src/ParseFile.js	96.84% <100%> (+0.12%)	⬆️

---

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 77e4068...74fd982. Read the comment docs.

[stevestencil]

I noticed that when saving the files by using request() (instead of ajax()) it sets the session token as "_SessionToken" in the body. With this change you're setting the session token in the header X-Parse-Session-Token. Unless I'm missing something, shouldn't these use one or the other?

[dplewis]

Looking here Both of these have existed since the beginning of parse. One represents the REST API with headers the other represents POST request with (headers included in data). In this case SaveFile is using REST style request so for consistency I'm using REST style header here.

I'm not really sure why they used ajax() instead of request() here, maybe there was a side-effect. We can try it in a separate PR. @acinader Thoughts?

For now does this work for you?

[acinader]

@dplewis this change looks good to me. I'd prefer to keep what you've done with the session in the header where we now expect it and wouldn't make a requirement for adding this important capability cleaning up a log established inconsistency (unless you want to that is).