On 10/17/2011 1:03 AM, Ronald S. Bultje wrote:
> This is great work and we should merge it (as in: patch LGTM).
> 
> Now, as for "explode", the distinction between "explode" and 
> "bitstream" appears to be mostly one of "stuff that used to be
> there before you touched it" and "stuff you touched". Not that I
> mind, but I feel that ATM, bitstream and explode do approximately
> the same thing. If they do, we should either split it cleaner, or
> merge it?
> 
> Ronald

There's indeed currently no clean conceptual distinction between
"explode" and "bitstream", as you observe. I also agree that this is
a problem. I wrote it that way so I could relatively quickly get
lots of new checks in without having to debate the status of each
one but that's not a good long-term situation.

Further, a nontrivial number of people might actually use
"compliant"/"crccheck"+"bitstream". Changing its behavior thus would
become dependent on either risking disrupting that by either
removing relied-upon checks or making it suddenly too strict by
existing "compliant" standards.

That risk stipulated, fixing up the "bitstream"/"explode" confusion
something less contingent on who happened to notice the error
conditions would make sense. It makes more sense to do so, though,
with the current (basically behavior-invariant) patch committed as a
base and by modifying particular err_recognition checks. It's no
worse than the status quo and allows for more flexible testing.

Also, regardless, the first patch,
<http://patchwork.libav.org/patch/10264/>, would remain unchanged
and could be pushed.

-Dustin
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to