katharine / httpebble-watch Goto Github PK

Much easier said than done, but perhaps an idea....

It would be nice to some how transparently (ish) split oversize http responses into multiple messages.

I'd love to take it on myself, but I'm a beginner to pebble so thought I'd post the idea in case other people wanted to work on it.

Perhaps using a similar concept to keep reading from the mobile phone until all data is read. I can picture this in c# similar to reading each line of a file until it's finished.... Of course in this context it would be very different - but as an analogy.

Keep up the good work.

Found this great tip here:

you can send a ['d', 'base64-encoded-string'] tuple from the server in the same style as the width-specified integers, and that'll be sent as the decoded binary data in a single key

Posting here for reference, and because it should be documented for any others implementing your protocol.

Hi,

I think my buffer in bytes is matching the max I'd expect from the server (by getting the number of bytes for a string and plenty more for headers etc). But I'm still getting HTTP_BUFFER_OVERFLOW.

What sort of thing should I watch out for?

Do I need to somehow clear the buffer after each request?

Thanks for the help.

Would it be possible in a future (major) version to drop the "unsigned 8-bit boolean (true/false) indicating whether the request was successful" and just communicate that via HTTP_STATUS_KEY i.e. 0 if unsuccessful and the server code if it was? This would free up 8 bytes for user data, as each tuple takes 7 bytes plus the value's size.

I'm getting a failure callback and an http_status of 1022 when I perform an http_location_request inside my app. As best as I can tell, the 1002 doesn't fit any of the error code constants. Do you have any thoughts on the matter?