improve signature memory efficiency

[drahnr]

trafficstars

For large rpms, currently the content is copied. For rpms with sizes beyond 100mb (the ones with many assets i.e. games) might suffer OOM conditions when running on tighter memory bounds.

Proposal: Migrate to a trait Read based approach, using a custom cursor which allows sequential read across multiple buffers (header + content).

Relevant loc: https://github.com/Richterrettich/rpm-rs/blob/master/src/rpm.rs#L70-L72 https://github.com/Richterrettich/rpm-rs/blob/master/src/rpm.rs#L1454-L1456

[Richterrettich]

Yes, I think a reader based approach would be nice. Do you want to tackle this?

[drahnr]

Yep, I should find a few hours to impl that plus do a repo structure refactor :)

[Richterrettich]

Nice, but please do the restructuring in a separate PR since it is easier to review then.

[drahnr]

I'll look into the cursor impl once #12 is merged.