Skip to content

Instantly share code, notes, and snippets.

@deobald
Created November 17, 2023 16:41
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save deobald/707a0a646bde869fe62701d48f5132fc to your computer and use it in GitHub Desktop.
Save deobald/707a0a646bde869fe62701d48f5132fc to your computer and use it in GitHub Desktop.
Endatabas Why Document - notes

Unchanged

  • "many-tentacled leviathan" - again, debate around the overall tone.
    • Too Lovecraftian? I agree this is too colourful but I'm actually struggling to write this sentence in neutral language without it reading so watered down as to feel like boilerplate, but that's a problem with the entire document, I suppose. Suggestions very welcome.

Fixed, but require sanity-check

  • "When considering alternatives, there are no contenders." - obvious/direct contenders?
    • I'm tempted to leave this, rather than hedge it. "Contender", when used in the sense of being in-the-running, definitely excludes all alternatives (for Endb), I think. An "indirect contender" is playing a different game. An "unobvious contender" implies we haven't researched the space. That said, this is hardly a hill I'd die on.
  • "each feels a bit like Adriano Celentano singing in synthetic American English. " - would drop this, as it's a bit dismissive.
    • Dropped. Are you okay with "suffer" in the previous sentence? I'm on the fence about it, but I do want to address the not-quite-SQL language category directly, and why Endb doesn't choose that route.
  • "One can fantasize about designing a query language from scratch but not only is this a lifelong endeavour, it's very easy to get wrong." - on one hand, original SQL (like JS) was designed pretty quickly, but then grew. The real problem is more building mindshare that's the lifelong part of this.
    • I added the mindshare part, rather than dropping the "easy to get wrong" part, but let me know if you think that's disingenuous? I followed with a sentence about SQL's journey, but I'd rather drop it if you don't think it's necessary.
  • "zip files" - unsure about this one. POSIX or Unix also works here.
    • Yeah, I originally had POSIX. "Zip files" was intentionally out-of-band but I struggled to come up with another example that wasn't another file format or another protocol. If you can think of another example, at another layer of computing, I'd put it back. ;) Put POSIX back for now.
  • "But we believe people will not buy new proprietary data products, since we certainly wouldn't." - again, a bit of tone.
    • I need a tone check on the replacement. To me, the new sentence reads as positive and hopeful. But I can imagine how it could be read as tired and curmudgeonly. ;)

Done:

Missing link on Appendix page. "Dramatic change" - maybe a better word? "A window of time has recently opened when all of this is finally possible, together." - maybe a bit hyperbole? "adhered to by no single database" - maybe mention that MimerSQL are aiming (unsure how up-to-date they are though). "If a company is particularly broken, it will extract metrics from logs to create invoices and reports." - again, a bit of the tone. "For every Postgres operator who's ever fought with repmgr, we want Endatabas to be seamless by comparison." - same here. "all offer some way of shoehorning nested data into flat tables" - some of them do support this natively, like PartiQL/Couchbase and Mongo etc. A bit of a tangent, but there's the entire world of -actual- document stores, XML, XQuery etc. "are run on a copy of most production databases today anyway" - I would tone down these kind of blanket statements I feel. "Many financial firms began building their own temporal databases around 2020." - feels like we don't know this for sure, contradicts earlier statement about Finance always needing audit a bit. Isn't GS Slang a temporal store of sorts? Seen various history and audit trails at all banks I've been. Not necessarily bespoke DBMSs though. ... change the year (vaguer) "It cannot be constructed from components of SQLite" - this isn't strictly true, you get different trade-offs and could imagine SQLite being the runtime similar to how Lisp is, but with a more mixed role (SQL backend, storage, execution).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment