Hector Martin

People love to get all excited about leaks of official SDKs or code... but really, rarely are they all that amazing. Early on in development of homebrew for a console, they can help make things move much faster, but at the expense of legally tainting everything they touch.

This late in the game, though? Some of my reverse engineered documentation is *more accurate* than the official docs. Because I document what the hardware *does*, not what they *intended* it to do.

So for example, the leaked docs talk about registers existing in certain memory windows, but don't document all the mirrors that I discovered back when I did an exhaustive analysis of the Starlet 32-bit address space.

Now, certainly, leaks like this do clarify a lot of the long tail of things. "Oooh, so *that's* what that register did!". But those things we didn't know are largely because they weren't important. If they were, we would have figured them out.

And it's always fun to see the official names for things. Remember the STM release exploit? That bug is in the leaked code. The real code has the same structure as my decompiled version, and the bug is exactly what I thought (missing return statement), but the names differ.