@jmoiron @phenlix @inowland It was v1/v2 message incompatibilities between gogo and the main line proto implementation, both Go implementations. The way it manifested was non-obvious and took a while to trace to what is a known issue (and gogo has very recently been officially deprecated...
jmoiron plays the blues
@jmoiron @LastonSlappy 🤦 hopefully what i meant to communicate (that people with agency did this actively, it didn't just happen) is clear
@jmoiron @inowland Yeah, I think we agree on that. My rant is that they've been successfully marketed and sold as starter tech choices (you'll need it eventually!), and people who helped make that happen are now saying "woo boy things sure got complicated!"
@jmoiron @inowland Great example, because we recently spent >3 weeks debugging a nuanced incompatibility between gRPC implementations. At 1000 ppl, I'd agree that the benefits of the ecosystem outweigh the complexity costs, but do you want to go on this side quest at 10 people? 50?
@jmoiron @inowland It's easy to say, but is it true? I'd say that the population
& influence of x-faang vastly outweighs "gods of pre-IPO", especially at large. What industry wide trends do you attribute to that population?
@jmoiron This has manifested mostly as a knack for building the right sized thing, and I wouldn't build our current tech for the company we were back then. That sounds ridiculous to me. And that's what we've done as an industry by defaulting to all these internet scale technologies.
@jmoiron As someone who played a significant engineering role at a company from full-co single-room standups to a successful IPO, I've learned a lot about scaling an org, scaling a tech stack, and the requirements of a company at many different stages of development.
@jmoiron This widespread adoption is good for big tech cos, because their scale required them to operate in a completely different technological universe from the rest of the industry. How can you bridge that gap? Export your engineering culture as "the way."
@jmoiron To paraphrase an old one from @benji_smith:
"Nobody really buys hammers anymore, they're kind of obselete... What you need is a hammer factory."
https://t.co/SGwUpfYKBk
"Nobody really does platforms anymore, what you need is a platform platform."
https://t.co/s8mcTkh5n2
@jmoiron We shouldn't passive voice "tech became complex", it was made that way on purpose. Loads of X-faang + conference circuit + devrel convinced everyone that YGNI, and with the smart money bought in, everyone else jumped board.
@jmoiron @elithrar tbh black sesame milk sounds like it would be delicious 🤔
@jmoiron @elithrar what, you don't want black milk?
@jmoiron @RayS @BenThePCGuy @Microsoft Consolas
@jmoiron In this new starship troopers game, you play as the fascists? Is this a "spec ops:the line" situation?
@jmoiron @JoelBarciauskas well, serious answer, if it really collapses, one or more things will fill that niche and i'll probably be on at least one of them :)
@jmoiron @RayS It's where millenials go to find their pogs or whatever
@jmoiron If things go south, I am incredibly googleable.
@jmoiron @phil_eaton "If you’re thinking without writing, you only think you’re thinking." https://t.co/8kxqWOnWuM
@jmoiron When IBM, HP et. al tried to sell us on CORBA/UML/SOAP/WSDL etc in the early 2000s, we were like "lol nope" and built everything with REST+json, but now FAANG evangelize internal practices that scale down poorly to small orgs and people think "yes this is the right way." https://...
@jmoiron @copyconstruct Everyone wants to know what differentiates staff/principal engs, and an huge amount of it is "good engineers who also understand how people and organizations work."
@jmoiron @simonw @oneunderscore__ Makes this aspect of NFT profile pics seem even more troublesome, since verification status was the backstop. https://t.co/Hd4wpzepCw
@jmoiron This explains why Google has such a great reputation for product development compared to Apple. https://t.co/X70Qy1PNch
@jmoiron As an Engineer, I know that every time you look at an unfamiliar but sophisticated system, your first reaction is "this is 10x more complex than it needs to be, we could get rid of half of it", and I can tell you that this reaction is always wrong.
@jmoiron @copyconstruct Other than thinking this was obvious based on my timeline, my first thoughts were that they missed a favorable IPO market 3-4y ago when they were clearly "ready" if they wanted to raise, and not being sure if that would have made this better or worse.
@jmoiron @MarcJBrooker No, this is exactly it. A lot of backends can handle some nominal shuffling in the temporal order semantically but are actually relying on this order for throughput, eg. to run compactions less frequently.