Base64 is bulky!

You will be able to get much more data stuffed into facebook chat if you ditch Base64 for Unicode, we also started with Base64, but IIRC it was a ~30% inflation of data size.

Check out this project I was a part of in 2009, we used Twitter as file storage system, so, pretty similar.
https://www.youtube.com/watch?v=kUcLbmU5fHk

Love the project, good luck!

[matiasinsaurralde]

Thanks for the info, I'll take a look!

The basic idea is you take your data blob, and encode it directly as unicode characters.

I uploaded some old proof of concept code, if you're interested.
There's a python version and C++ version of the datablob -> UTF8 encoding function, but you may have to dig around for the good stuff.
https://github.com/C-P-949/TwatFS