On 08/15/2017 08:00 PM, Peter Silva wrote:
Isn't there kind of a universal issue that tar and compression happen sort of in the wrong order? Wouldn't it make more sense to make files that were .gz.tar (ie. compress the files individually, then have an index into them via tar.)
That way gives a lower compression ratio, especially when you have a bunch small, similar files. E.g., something that ships a lot of C header files. Or a bunch of plain text files.
Also of course there don't exist any good tools to create or extract such archives, but that could be fixed (afio exists, but it's non-free).
Also—if I'm remembering correctly, tar has no table of contents, I think listing a tar file is done by reading all the individual file headers, which are spread throughout the file.