Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Agreed, and this is actually a really good point:

> we don't implement a feature unless it works for the general cas

I used to make that mistake a lot. My boss would say "can we do this", for example report memory usage per operation. And I would say "no because sometimes memory is shared between operations so it would be meaningless". In other words I couldn't do it perfectly so I said we couldn't do it at all.

That's what the VLC guy is doing and I didn't realise until I worked with that boss that it is COMPLETELY WRONG!

Just because you can't do it perfectly doesn't mean giving up entirely is the best you can do. In cases like this you can absolutely do something that works sometimes but not in every case and that is way better for users than just giving up.

Lots of programmers fall into that trap though.



VLC is able to abstract over a ton of extremely complicated codecs by providing some common tools that work for all of them.

I guess if VLC has a feature, you can always expect it to work. That's their design choice. There's nothing "COMPLETELY WRONG" about that.


The most entertaining part of this thread is that VLC is actually in the process of replacing its backend with the mpv-derived libplacebo.

In the end, VLC has admitted defeat


Both use FFMPEG anyway so it's not like VLC's codec abstraction was really something that sets it apart.


Exactly an in this case it even works 99.99% and VLC already has a related feature (seeking to a specific time) that has pretty much the same codec requirements.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: