I went to a talk on YouTube. It was about how they have been
adapting to changing demands over the years. One of the major
issues was the decoding/encoding of videos (including enhancement
of quality). For this they break the videos up into digestible
chunks and encode/convert/edit the chunks separately. At the end
of the process they put all the chunks back together again. A
video can thus be processed in 5-10 minutes.
The second major point was, of course, streaming and buffering.
I was thinking that D with ranges and component programming might
be of interest for YouTube. The only thing I was worried about
was the memory bit. The demands are quite tough (low footprint,
speed). I know that GC can be deactivated in D but the library
still depends on it. In my projects GC does not hurt, but for
life streaming and the like it might be a different story. Does
anyone have any experience with D in this respect? Any opinions?
