Re: [Gnash-commit] gnash ChangeLog libamf/amf.cpp libamf/amf.h lib...

[Rob Savoye]

strk wrote:

> What was wrong in checking input boundaries before reading ?
> We'd be back to memory corruption instead of exception throwing
> in the malformed-or-tought-so case...

  It was obviously added somewhat adhoc, and not really thought out.
Instead I plan to add support to my test cases (probably next week) for
randomizing packets similar to zzuf, and then using that to make the
code a bit more robust. That I'll be trapping real errors, instead of
just guessing. That should keep us both happy. Rather than checking
after every few lines of code, I prefer testing at a higher level
anyway. I did add some other tests as well for the buffer. As almost
nobody is even using all this code, I don't think there is a big
exposure over the next few weeks, and it'll all be rock solid by the
next release. (after 0.8.3)

        - rob -