Why is the header text-based?

[DavidBuchanan314]

From my perspective, the decision to use a text-based header format is a slightly strange one.

As an implementer, extra care must be taken regarding whitespace handling, ensuring that base64 encodings are canonical, etc. In general, text-based parsing is more fiddly and has more edge-cases to consider than binary format parsing (IMHO).

As an end-user, there is little benefit to being able to read the headers visually, since they're still very opaque. Information about the number and types of recipients could easily be reported by a command-line tool option, if desired.

Is there some benefit or rationale that I haven't considered?

During my process of understanding age, I've uncovered a couple of issues in existing implementations, both of which I'd put under the category of "parsing text is hard":

• scrypt recipient parser accepts non-canonical integer encodings jojonas/pyage#6

• Quadratic time complexity when parsing long recipient stanzas str4d/rage#368

I can't help thinking that these issues could've been avoided if age used a binary format. Maybe something to consider for an "age v2"?

(note: I originally created an issue here, but I've now moved it to this discussion since this seems to be the best place for it)

[DavidBuchanan314]

I've found this discussion on the mailing list which mostly answers my question https://groups.google.com/g/age-dev/c/Td-2MtU6p1I

I agree that header compactness should be a non-goal, but my interests lie in making parsing easier and safer.

Being able to look at a file and easily see what it is

You can do this with binary formats too, if you make the "magic bytes" contain a human-readable string. (better yet, have those bytes added to the file database)

I picked text because I find it easier to work with in modern languages, and existing binary formats come with a lot of baggage.

I can agree on the second part, but I can't agree with the first - although that's clearly a matter of personal opinion.

I agree that modern languages make dealing with strings nicer, but the flip-side is that older languages (e.g. C) make dealing with strings a pain. I believe that a binary format would be easier to implement across a diverse set of languages.