On Fri, Oct 3, 2014 at 1:37 PM, John Hudson <tiro@tiro.com> wrote:
> Jonathan,
>
> Out of interest, are you writing your own Brotli decoder, or using open
> source code from Google?
Using Google one I believe.
> I wonder because some of my colleagues have expressed concern about the
> single implementation of Brotli compression/decompression, as compared with
> the variety of libraries available for handling WOFF1 gzip. They wonder how
> robustly tested is the Google code, given how new Brotli is? Also, has it
> undergone security reviews?
Those are very valid concerns. In fact, when I reviewed the code before I
found a few alarming issues. In particular, for Brotli itself. Granted,
this might not be exploitable currently, but definitely should be addressed:
https://code.google.com/p/font-compression-reference/issues/detail?id=2
Possibly affecting the metadata compression:
https://code.google.com/p/font-compression-reference/issues/detail?id=4
And a more generic issue in the woff2 implementation:
https://code.google.com/p/font-compression-reference/issues/list
Now might be a good time to address those.
behdad