thdxr

The answers here focus a lot on improving collaboration. Almost all challenges in running a company come to this and there definitely is a lot you can do there. But it's hard and few companies pull this of and no one really wants to admit it. You usually get vague suggestions like "better communication" or "work less in isolation"

I've taken a different approach and I have become extremely conservative in extracting out responsibilities in a dedicated role.

I no longer try to build teams with the design role externalized in someone else, I just focus on finding developers who are also good designers. This is hard but I'd rather fight this battle than the collaboration one

If there truly is a 10x designer that wants to work with me then it's worth it, otherwise it's synergy of having one person be able to fully ship on their own outweighs the overhead of collaboration

The issue I ran into with this was I could initiate a login for anyone and if they accidentally hit the link I could remote login.

Switched to pin code entry, less fancy but achieves all the desired UX

Is there some kind of standard / open source specification for "formulas"

I see this concept in a few applications and wondering I'd need to re-invent my own

it was exactly this kind of thinking that lead to Japanese automakers crushing American ones

vendor relationships do not have to be hostile. AWS has been around for a while and has never raised prices, some services have gotten 99% cheaper.

yes there's some hypothetical day where they say "tricked you, all that stuff about margin being opportunity was a lie" and they try to extract value short term while burning their reputation

but every day that goes by where this doesn't happen is $$$ wasted on trying to avoid it

There are a lot of tools in AWS to deal with this. Api Gateway Rate Limiting, WAF, etc.

That said it does take a level of awareness to set these things up. Some of what we try to do at SST is turn on these for you automatically so you're not being punished for not knowing something

Whenever it comes to things like legal issues or regulations, people take such pride in being someone who understands their intricacies that the larger picture is missed. We should think about what outcome we actually want.

While technically any snippet of code can claim a copyright, it's sort of ridiculous to suggest the snippets that Copilot generates has any protectable value.

Practically Copilot is saving you the labor of writing fairly trivial but tedious code. I work on open source full time, there's no random snippet of code that I've written I'd feel upset about if someone copied.

Licenses on software are mostly performative.

there is a group formed including Deno that will work together to try to maintain standards + compatibility

https://blog.cloudflare.com/introducing-the-wintercg/

This has been the biggest productivity improvement to my workflow in years

No it doesn't "understand what I'm doing" or "get everything right" but that's hardly the point

It's often reducing the amount of labor I'm doing by hitting the keyboard by guessing 90% correctly what I was going to type

It also often saves me from having to google how to do something, it's effectively serving me a search result right along my code

I'm lucky to be getting it for free but would have immediately paid $10. It needs to only save you minutes a month for that to be worth it

Also the comments about it being "unfair their monetizing other people's work" are missing the point.

Github has created a product that many people use and through that effort created a large repository of code.

They are now releasing a product that is going to create a large amount of of value in time saved and are maybe capturing 2% of that. This is a great outcome for everyone

HN is a mature product! Most common software failures are from a deployment of changes, I don't think HN is being deployed to every day

We're all used to effectively beta software that's constantly being updated every day and never final

Very interesting!

Note lambdas now can have urls if you want to set that up without setting up API Gateway

Looks like the other rough edges are from being understandably overwhelmed by AWS. Would love to chat with the author of this article about how we can make WASM an option in SST (https://sst.dev) so this is becomes a much easier option

what's funny is firecracker, the underlying tech fly is using, was created by AWS

only used by Lambda at the moment though, not in ec2

been pairing DynamoDB with rockset to solve this

I've seen incredibly long prompts to DALL-E to get a precise and incredible result, much better than anything I could come up with

like everything before it, it's just another tool that some will show a 100x mastery of than the average person

in practice how do you make a single application node the writer?

do you now need your nodes to be clustered + electing a leader and shipping writes there?

know fly.io did this with PG + Elixir but BEAM makes this type of stuff pretty easy

This sounded oddly similar to my Twitter experience and then I noticed we're more or less in the same Twitter bubble

I totally understand your experience on Twitter and a lot of the comments here reflect it.

However I do not relate to your experience of it. Over the years I've curated a nice set of people to follow and I've had extremely rewarding experiences. I've learned a lot, invested in companies, received investment, all stemming from interactions I've had on Twitter.

I definitely occasionally see some of what you're talking about and it creeps into my bubble but I'm diligent about tracking down how it got there and unfollowing whoever pushed it into my field of view.

If I can be frank your post here is an example of the kind of stuff I usually push away. Fairly absolute stance with little articulation/empathy of the opposing viewpoint. Taking some of the genuinely negative behavior and implying it must be the entirety of everyone using Twitter.

Twitter tends to reflect yourself and if you're honest with yourself you'll often see that you seek out the things that upset you.

They need to offer sane paid products to fund themselves properly. With their reach they don't really have to be too creative.

For example, they could sell more features on top of their password manager and compete with things like Bitwarden.

There are a dozens opportunities like this, the fact that they still go towards ads as a way to make money is what has me pessimistic about their management.

This is great work - thanks for putting this together. We recently got a request to add Rust support to ServerlessStack and looks like there's good reason to :)

Others have written about this better than I can, here's an article: https://dev.to/garethmcc/why-local-development-for-serverles...

Paul Swail has spoken a lot about this and has an upcoming book around the concept: https://serverlesstestinghandbook.com/

We test it against real AWS deployments. With a proper multi-account strategy + SSO, there are no AWS credentials to leak + blast radius is limited.

The feedback loop is definitely a problem but projects like SST return the instant feedback loop to you.

It's impressive what Localstack has been able to do to keep up with AWS.

However for my own projects, we consider local development like this an anti-pattern. IaC tools like Terraform/CDK/SST make it easy to spin up environments for each developer in the cloud. This may sound crazy at first but would recommend giving it a shot.

If you're fully taking advantage of managed AWS services the best development environment is running everything in AWS. There's definitely some friction but you reduce the "was working on my machine but not in production" situations.

Elixir and Erlang aren't async everything. Everything runs synchronously by default, you need to use separate primitives to turn it async

With how many banking startups that are showing up with fairly polished products, Google really reminding us how much they struggle to execute on anything

I'm really excited about this direction, a few years ago I asked the question "who will be brave enough to write JS tooling not in JS" and it seems a lot of projects now exist.

That said, I do not believe Rome will be the winner here. If they were willing to undertake this giant project initially _without_ the benefits of using a faster language, I'm not sure what the vision was.

Really excited about esbuild, swc and bunjs

This actually only builds and does not type-check similar to esbuild. The author is working on a proprietary type-checker in Rust though

Once you have to pay a price for your politics then you see what people really believe

Relational vs non-relation is one of the biggest misconceptions about NoSQL

Data is usually relational, there's no getting away from that just because of the database you use. NoSQL can manage all kinds of relations, just with an extremely different approach.

If you're interested in learning how this works instead of repeating the tired meme of "mongodb lol" checkout Alex Debrie's book on DynamoDB and Rick Houlihans YouTube content.

I also wrote a brief thread about this on Twitter: https://twitter.com/thdxr/status/1394903426023272452?s=19

Full stack serverless. It's always important to reduce ops but even more so as a team of 1.

I use serverless-stack with every AWS service I can possibly use to reduce my code liability.

Going through the Clerky process right now. It's OK not blown away. Used Gust Launch in the past and recall that being super easy

The overhead with this architecture can be cumbersome. Which is why most successful deployments of it tend to be with teams that have embraced full stack serverless. Recommend exploring that community, plenty of event driven systems there.