Implemented support for chunked transfer-encoding

[danielgindi]

This will handles issues like:
request/request#2896
request/request#2488
request/request#2380
request/request#1770

[coveralls]

Coverage increased (+0.2%) to 98.206% when pulling a7bc1c3 on danielgindi:feature/chunked into 0170178 on form-data:master.

[danielgindi]

AppVeyor fails due to permission issues

[BenjD90]

Is there any link with this issue (request/request#2499) ?

[alexindigo]

@danielgindi Sorry, somehow I missed it. I'm loving it. Let me try to run it locally for sanity check and I'll check what's up with AppWeird.

Thanks a lot. It looks awesome.

[alexindigo]

Seems like it's npm@2 issue. Let's update node version (both appveyor and travis) to 6, 8, 10.

[alexindigo]

Looks like that whole "NaN as chunked" use case is missing

Do you mind to update tests to cover that?

[alexindigo]

I think that and bumping node versions will bring us to the finish line.
Then we'll cut new major version and everybody will be happy. :)

Thank you.

[danielgindi]

No problem, I found out that that uncovered line will indeed never happen in normal conditions.
It could only happen if a "familiar" (readable file, http request...) stream is supplied and then its properties will be changed just before calling getLength.
Since the library is not responsible for handling such race conditions, and this would be an unrecoverable situation anyway - I removed that redundant line.

[danielgindi]

And I'll let you bump the version :-)

[danielgindi]

@alexindigo any news? currently having to do hacks in code to support this...

[naugtur]

For anyone who ends up here from issues with multipat where you wish to set encoding to chunked.

To prevent request from setting content-length on the request without any changes to it and this library you must set
knownLength: NaN

That will explicitly propagate through length counting and cause resulting total length to be NaN thus stopping request from setting content-length.

If properly documented this might be considered a feature ;)

[danielgindi]

@naugtur This is a very nice workaround :-)

[Exegetech]

@naugtur you just saved me....