@feross@vladsazonau Nice how do you do network proxy mode to avoid bypass? What mdm has been most successful in your opinion in conjunction with socket? Open to dm if you’d prefer to connect there.
@feross@vladsazonau How do you enforce this for the team and ensure you never not use it? Can it be deployed over mdm and ensure the old npm install is never accidentally used?
@dangtony98 How do you do dynamic secrets for a database like rds mysql/pg (apart from the fact that iam roles are better suited here)? Are you actually updating the password every few minutes with a new GRANT.
I'm a bit surprised about this clause disallowing benchmarking of Planetscale https://t.co/o42AFfdfBJ - access any portion of the Services for the purpose of building a similar or competitive product or service, or monitor the Services for any benchmarking or competitive purpose;
@PlanetScale@samlambert Am I missing something here? Benchmarking seems to be encouraged here: https://t.co/cPuNyLvYlQ but disallowed in the AUP
@towheretobegin@willchen500 Agreed there’s no risk differentiation unless a firm becomes risk bearing, and is willing to bear those costs (which could be enormous), so it just boils down to the consulting model, which is what it was mostly before.
@willchen500 Also most importantly, who is risk bearing, will the vendor take on liability for mistakes. Open source software generally bears 0 risk, which is good since at least the line is this clear. The alternatives are murkier on this line which is why they are suboptimal.
@willchen500 This seems like the right direction, but I’d imagine most of the value is not in the software but in the evals on a firms own test sets and the implementation to curate those: did it come up with the correct results in these cases, etc.
I'm a bit surprised about this clause disallowing benchmarking of Planetscale https://t.co/o42AFfdfBJ - access any portion of the Services for the purpose of building a similar or competitive product or service, or monitor the Services for any benchmarking or competitive purpose;
@PlanetScale@samlambert Am I missing something here? Benchmarking seems to be encouraged here: https://t.co/cPuNyLvYlQ but disallowed in the AUP
@NathanFlurry New software products touching user data that isn’t open source and can’t run fully dedicated in customer cloud is feeling DOA right now. Customers are taking on way too much risk with vendor sprawl and lack of their own ingress/egress controls.
I'm quite impressed by S3 files. Specifically that we don't need S3 GET's and PUT's everywhere in our code any longer, the code can just read and write files locally, and in the cloud the same but via a S3 mount point - More thoughts here: https://t.co/7sKKvBIj3p
@brankopetric00 The policy framework had terrible dx but incredible control if you could figure it out. Claude is now giving the engine the dx to be wielded effectively.
@ankrgyl Great read. Did you also your have thoughts on clickhouse and data fusion? A lot of folks in tracing are using clickhouse but others like logfire also built on top of data fusion. Good to learn about tantivy.
We run it hosted in our vpc so anyone on the team can push to it vs running a sidecar locally when testing their apps, and there’s a reset button in the ui to truncate all data quickly. Great dev experience to see your traces logs and metrics fast and easy for anyone. Only thing is the default port for grpc is 18889 and ui is 18888 but easy to remap these of course to 4317 and 8888 if you want.
@thdxr aspire dev is quite good, much better than jaeger for metrics logs (structured or not) and traces. all in memory and has mcp as well and other bells and whistles. Run it with DOTNET_DASHBOARD_UNSECURED_ALLOW_ANONYMOUS=true for no auth needed