[kevin_thibedeau]

Tar is a convenient format for basic embedded data storage since it's easy to parse and you don't have to build custom tooling to prepare them. We'd store small pre-gzipped web resources in a tar file and send them out as is without needing onboard compression. The blocking factor puzzled me for a bit when I was trying to store a few hundred bytes and couldn't understand where the extra 10k was coming from. Sometimes you have to RTFM.

[bestouff]

I wrote an incomplete tar parser (genext2fs author here) because I also thought tar was easy. Believe me nothing could be more wrong. That old underspecified file format is full of weird addons by different implementations, bug workarounds, it's big and hard to get right.

Eventually I gave in and added libarchive as an alternative.

[kevin_thibedeau]

If you stick to the modern format it's simple. If you are only using modern tooling there's no need to support the legacy headers.