Convert the code to Rust about notqmail

[josuah]

Refactoring the code would also break all the 3rd-party patches there are for qmail.

If qmail gets widely acknowledged by the community and we end-up with non-patch extension mechanisms (already widely supported by qmail by adding or replacing one of its programs), followed by most, then I guess this removes the need of having a pure-C codebase.

In the other hand, if there are such a project of rewriting qmail in another language, it will be interesting to know it for notqmail team, so that all initiatives remain compatible: the mail queue format, install locations, programs naming and pipe/file descriptor interaction.

[schmonz]

Thank you for this well-structured suggestion, including rationale, tools to help with incremental migration, and a worked real-world example. I personally find the idea more than a little intriguing (perhaps especially because it would provide me an excuse to finally learn Rust).

The biggest tradeoff I can think of is, as soon as any important part of notqmail is written in Rust, we lose a great deal of OS and architecture portability that has historically been one of qmail's strengths. This might well be a worthwhile tradeoff, given how risky it is to live with C. No matter what we decide, it will be a consequential decision.

[najamelan]

@schmonz You can find the supported targets/architectures here.