3 ms·
I think a lot of this is based on the trends in the industry at the time that no longer necessarily hold true. Once you get past the ways to encode bits and byt
by int_19h 3y ago
I think a lot of this is based on the trends in the industry at the time that no longer necessarily hold true. Once you get past the ways to encode bits and bytes, the stack has been remarkably stable if anything. E.g. PNG is still the most common lossless bitmap image format and there's no risk of support for it being dropped anytime soon - and it's over 25 years old at this point. Similarly, there was a big mess with text encodings, but now Unicode is pervasive and is not going anywhere anytime soon. We also have large archives of information on how to interpret all kinds data in even larger archives (e.g. Wikipedia has lots of details on specific file formats, even fairly obscure ones). I'm actually fairly confident that anything that is actually archived today would remain readable for a very long time to come.
One thing that could possibly be done better, though, is self-describing formats. I don't think it would require that much extra space to basically attach a terse but unambiguous description of how to interpret the data to most if not all binary container formats out there. Something that has minimal encoding, so much so that doing the equivalent of "strings" on the file would be sufficient to see it.