Get callback get fired twice about needle

[tomas]

Holy shmoly. I'll take a look at this ASAP.

[tomas]

I think the problem lies with the server responding with a Content-Length of 0 bytes. That makes the parse error occur at the socket level, not at the JSON.parse() level:

Error: Parse Error
    at Socket.socketOnData (http.js:1583:20)
    at TCP.onread (net.js:527:27)

I'm not sure how we should handle this. I could skip the error if its code equals 'HPE_INVALID_CONSTANT', and you would get an empty buffer as a response. What do you think?

[a70ma]

Personally i don't mind getting an error ( any error ) while i won't get the second callback it's fine.
But i'm not a fan of hardcoding stuff, better build a simple mechanism to keep track when a callback was fired ( like making callback = null or fired = true etc ) , because i'm not sure if HPE_INVALID_CONSTANT is the only case.
In the web scrapers world things are scary.

[tomas]

Yeah, I know (been there). Needle's point is to make it easy for developers, so I guess we should handle this case in a better way. After all, what everyone cares for is the data, not if the data's full length matches what the Content-Length header says.

[tomas]

On second thought, and after reading about the HPE_INVALID_CONSTANT error, we're not handling that error code because it may occur for other reasons as well, such as the response not having an HTTP header. I think it's too dangerous to not trigger this error if the HTTP parser raises it.

I just pushed an updated version that does check whether the callback was fired or not, though, so you shouldn't get the callback fired twice now. I'm closing this issue but feel free to reopen it if you spot any other weirdness. :)

[a70ma]

Thanks, i'll do that tomorrow.