@david_whitney@davidfowl@JamesNK The JVM world has the gradle build system whose definitions are written in Kotlin. It can indeed get a little mind bending
@DalSoft I have spent the last few years building out a platform around Kubernetes, and don't recommend it for a lot of use-cases. I think you're close here: it is an amazing technology but a single project is rarely a good candidate because the required investment isn't trivial
@CFDevelop I noticed my timeline became really horrendous last week before I realised I accidentally opened "for you" instead of "your timeline". Just aweful
@ICooper@MartinDotNet It was as specific as I could write whilst making an emergency cheese sauce, but yeah, inter-bounded-context is probably links, intra-bc is /probably?/ the same span? This talk could provide valuable insight
@MartinDotNet@ICooper I think links are pretty underserved with respect to documentation. Their how and why would both be useful because most o11y practitioners don't seem to have a grasp on them
@ThisIsJoFrank Using the first random links as an example of each:
Azure's IaC documentation e.g.
https://t.co/1NrIPFBoFU
And Envoy documentation come to mind e.g.
https://t.co/e6dVXCNRqI
They are both pretty exhaustive and link to human-curated examples
@MartinDotNet (Prometheus) Counters may only go up. If the exposed value goes down, it causes chaos with the metrics. I guess something that goes either way would be a gauge?
@david_whitney Not sure about this. Standardising on k8s allows us to create common ci/cd for many apps, canary rollouts, alerting and other automations which work on-prem and on-cloud. Still use our cloud's bespoke features, but runtime hosting is less interesting.K8s=CLR; cloud=nuget packages
Hi @programmingart, thanks for your http/3 series! Why is the Connection Identifier unencrypted? I'm guessing it won't work well for sticky sessions / sharding because it can change, but I can't think of another use-case
@sterenas@rbranson@kellabyte Its true that the amount of space taken by a larger varchar is the same, but there still may be costs. E.g. some dB engines will allocate memory when querying according to the max length of a column, regardless of the data actually stored
@housecor Pretty obvious TBH
Network cables heat up as you pass more data; the first records are raw and dangerous, and everything after is too burned to be useful
@DenisTRUFFAUT@mhevery Bloom filters aren't resizable, and you use multiple hash functions concurrently. You do some math stating how many total items you need to support, how many hash functions you're using and the rate of false positives you can tolerate to get the correct size.
@code_is_@mhevery They both use hashing, but are quite different:
* BF are fixed size
* They are backed by a bit-array & don't store the input
* One value can set multiple bits to true
* Multiple values can set the same backing bits true (no collision avoidance) hence probabilistic behaviour
@david_whitney@EE I use https://t.co/seWBg4njVM - they are actually 3 (not even an mvno), no price rises, no roaming fees and ยฃ10 for 30GB
Only downside is no Wifi in the tube