On 6 January 2016 at 15:14, Ben Woolley <tauto...@gmail.com> wrote: > I know the proposal is "modest", but the timing can be represented in file > names that are microseconds from start, with an optional dash followed by the > microsecond to jump to (for looping). > > The main issue with this is that all file names would need to be read to > project the effective time with loops considered, which I think with tar > would require scanning the whole archive. In that case, a meta file would be > better. > > Sent from my iPhone > >> On Jan 6, 2016, at 11:55 AM, Leander S. Harding <l...@lsh.io> wrote: >> >>> On Wed, Jan 6, 2016 at 10:45 AM, FRIGN <d...@frign.de> wrote: >>> Hello fellow hackers, >>> >>> I'm pleased to announce version 1 of farbfeld[0]... >> >> Modest proposal: let us define an animated-farbfeld format as a >> tarball of sequentially-name farbfeld files, perhaps with a metadata >> file containing frame times and looping information. >>
That's an issue with tar. Tar is limited in it's design, probably because of the physical medium it was designed for [0]. There are other formats that can handle this type of problem, and problems similar to this random access issue in a more efficient manner. I don't think any animation would amount to this being an appreciable problem though. Animations are not so large that I would think this is a problem, and to add a metadata file for the one use case of knowing the animation duration doesn't seem suckless. [0] http://duplicity.nongnu.org/new_format.html
