[Michuelnik]

While symmetric GPG encryption produces binary output the asymmetric GPG encryption uses ASCII armored output that are roughly 30% larger than the binary output.

This patch omits the ASCII armoring part.

I've built the patch on my forked branch and used it on my production server successfully:

• The tar.gz.gpg has roughly the size of the tar.gz
• is binary encoded and
• has been successfully decrypted and unpacked

No tests were added since nothing was added, the existing tests should cover the new behaviour in the same way it did before. I successfully ran your lint, integration and unit tests on my fork, too.
No docs were changed since I assume the previous internal behaviour has been just a minor bug.

[m90]

Thanks for working on this. Even if I'd say the current behavior is a bug, I'd still like to keep the option to produce armored output if needed (the bug is the inconsistent output, not the armored output itself. Also someone might have built a pipeline around the behavior by now). I'm fine with defaulting to binary output, but I wouldn't want to remove it entirely. Thanks.

[Michuelnik]

I see what I can do. I will add a configuration option GPG_OUTPUT=ascii that switches between binary and ASCII output for asymmetric GPG. Allthough switching to binary as default might be a breaking change I will use this as proposed by you.
Adding the feature for symmetric GPG is beyond my capabilities of golang.

[m90]

Allthough switching to binary as default might be a breaking change I will use this as proposed by you.

It's just as breaking as changing the output of asymmetric encryption from armored to binary without adding an option, no?

Adding the feature for symmetric GPG is beyond my capabilities of golang.

I'll merge your work into a feature branch and will pick up from there, thank you.