[FFmpeg-devel] [PATCH] movenc.c: fix metadata writing.

[Ronald S. Bultje]

Hi Reimar,

On Sat, Mar 5, 2011 at 7:59 AM, Reimar D?ffinger
<Reimar.Doeffinger at gmx.de> wrote:
> On Sat, Mar 05, 2011 at 06:52:00AM -0500, Ronald S. Bultje wrote:
>> On Sat, Mar 5, 2011 at 6:16 AM, Reimar D?ffinger
>> <Reimar.Doeffinger at gmx.de> wrote:
>> > On Fri, Mar 04, 2011 at 03:03:35PM -0500, Ronald S. Bultje wrote:
>> >> On Fri, Mar 4, 2011 at 2:05 PM, Reimar D?ffinger
>> >> <Reimar.Doeffinger at gmx.de> wrote:
>> >> > Fixing MKTAG still would be a rather good idea...
>> >>
>> >> I'm not sure if MKTAG() is broken. It doesn't specify signedness or
>> >> range of the input characters.
>> >
>> > Can you think of even a single case where the current behaviour
>> > makes even the slightest bit of sense?
>> > For me the answer is no, and thus it is broken.
>>
>> Are we still talking about MKTAG?
>>
>> I think MKTAG does the right thing. You generally call it with hex
>> values (0x80, which is positive) or ascii chars ('c', which is
>> positive), and in both cases the result is valid. The only case where
>> the result is invalid is when you literally feed it wrong input, e.g.
>> -1. But that's not really MKTAG's fault, is it? It could certainly be
>> made more dummy-proof, but again, is that really MKTAG's fault?
>
> Failing for no good reason at all is just idiotic, and code that
> does idiotic things for no good reason is wrong.

What is the idiot thing that it does? Do you want me to cast to
unsigned inside the macro?

Ronald