Looking for Feedback on my Binary Serialization Format

[dseller]

nope it isnt, in the gist there is already a revision history, and i also have some older versions of the document in my pc. i suspected this would happen anyways, if i were telling AI to do this i would ask the AI for the feedback as well, why would i bother to get real feedback from humans for an hallucinated format? it may be a shitty format and i could agree with that, but i spent a fair chunk of my time writing it, reorganizing it, changing single words so things become clearer, if you read it its pretty clear its not AI written anyways, i have the idea that you havent read the doc for more than 2 minutes

First, it's littered with em dashes. Second, the whole flow of the document just screams LLM to me. Third, it's just filled with garbage nonsense extra fluff that's really not how humans write a document about a binary serialization format. Fourth, the "it's X, not Y" phrases. Fifth, all this weird talk about "substrates", "gardeners", "plants", weird chapter names like "DTBN Implementation Floor", and also phrases like "In summary, the DTBN binary stored or transmitted must be compliant. Everything else is fair game." just really give it away. Who writes that in a serialization format spec?

Hey, you wanted feedback. You got it.

[SlimiestSlimy]

i mean, im not a specialist writing specs with 10+ years of experience in binary formats, so dont judge me for my poor profesional writing skills lmao
i did use AI to correct the document after i write on it(spelling, references ,example checks, too tedious work to do manuall, here is were some em dashes come from, i hate them so i remove them so i know you're lying on there unless one escaped, i do love normal dashes tho) then i do another pass of writing, then i bump the revision number arbitrarily.
"everything else is fair game" i literally wrote that, again, im not a profesional. what i want is to get the point across, not make a peer reviewed paper. furthermore, what do you even gain talking abput this insteas of giving feedback? welcome to the internet myself i guess

also im pretty sure AI would come with something simpler or more practical, you criticize my poor garden metaphor, it took like a whole day for me to come up with it because on previous internet interactions it was obvious people were misunderstanding the purpose, i even changed the name of the entire thing to reflect it better. and im not sure whats "AI" about substrate, the original word i used was plain old dirt, found it impractical for the metaphor

there are 11 "not"s in the document that is like 60k characters and 9000 words as per a generic word counter on google.

I have the impresion you hate AI so much you haven't even seen it enough to recognize it lmao.

[Octocontrabass]

it helps to understand the data by having the extra type expressiveness

You already need some other way to understand the data, which makes your "extra type expressiveness" completely redundant.

also im pretty sure AI would come with something simpler or more practical,

LLMs are designed to do what you tell them to do even if it doesn't make sense.

[SlimiestSlimy]

i mean, there's also the revision history on the gist, but thats just soft proof
and i think you fail to see all the possible use cases. forensic analysis or debugging of foreign data can benefit from the extra semantic meaning, and like, its just 2 bytes for a type id, a jump table in a fixed implementation can handle the different name but same thing type ids just fine, or just use uint and int everywhere, the document does not force you

[dseller]

i mean, there's also the revision history on the gist, but thats just soft proof
and i think you fail to see all the possible use cases. forensic analysis or debugging of foreign data can benefit from the extra semantic meaning, and like, its just 2 bytes for a type id, a jump table in a fixed implementation can handle the different name but same thing type ids just fine, or just use uint and int everywhere, the document does not force you

Okay, well, you wanted feedback. Multiple people gave feedback. You can choose to ignore it if you want.

[SlimiestSlimy]

im choosing to ignore the feedback because its irrelevant, what does it matter if AI had to do with it? and i would appreciate informed feedback, at least read a bit of the document, ive received better feedback om other places, but definitely not here, it seems like yall are primed for conflict or something

[Octocontrabass]

forensic analysis or debugging of foreign data can benefit from the extra semantic meaning

Field names, like in JSON, would be much more useful if you wanted semantic meaning to help with forensic analysis and debugging.

[SlimiestSlimy]

Indeed, but json wastes a lot of bytes
DTBN aims to be a middle ground, not to fully explain what each type is for

[skelnel]

I have no experience in designing a binary format of any kind, but in general with any concrete proposal I think it would be helpful to define a problem that is attempted to be solved, so it would be possible to evaluate the solution critically and constructively from that perspective. Otherwise it's probably expected to not get that sort of feedback. I only glanced the first page and there it's stated explicitly that it's "not to solve an existing problem" and any concrete problems I could think of as a novice such as "error handling and validation, both of any kind, are implementation-defined" i.e. out of scope.

A lot of emphasis appears to be on the typedness and descriptiveness, so perhaps the itch attempted to be scratched here is a specification language for some purpose. If I think what is the fundamental difference between a text and a binary format, it would be that text is inefficient because it needs to be human readable, and binary is allowed to be as efficient as wanted because it doesn't have that constraint. Therefore I guess I would define a specification format as text because it's for human consumption.

[SlimiestSlimy]

the spec itself is not made for human consumption if i understood correctly what you wrote, and i agree with you in that i could be clearer with the objective.
the objective is a self describing binary notation that i can implement incrementally and partially without breaking the spec itself, and since it's mostly for somewhat related personal projects i care about all of them storing their data on a common format. the other aspect i could made clearer is in-band vs out of band information necessary to read and interpret the binary notation, ANY format needs some type of out of band information, BTDN chooses a somewhat uncommon middle grpund where we try to be a little more descriptive with typeids, but not full on JSON

[Octocontrabass]

since it's mostly for somewhat related personal projects

Is it possible your intended use cases have influenced the design of the format? You asked us to consider it as a generic binary serialization format, but several of your choices don't make sense for a generic format.

[SlimiestSlimy]

entirely possible, but I personally think its actually pretty generic, there are many different ways of representing the same data, and a lot of dials implementations can tweak while staying compliant, i keep updating it every day, just now i revamped the first sections to try to make it more clear what TDBN is about, but im mid editing so the gist may currently have some glaring issues, half the examples have wrong typeids at the momwnt lmao, but its the last time im reorganizing the typeids sooo i can deal with it,

[dseller]

It's as if you are trying to find a problem to match your solution instead of the other way around

[SlimiestSlimy]

do you know fun is not necessarily practical right?
whats your point anyways?