Hope you're all doing okay out there ๐
29.07.2025 22:56 โ ๐ 2 ๐ 0 ๐ฌ 0 ๐ 0@nblumhardt.com.bsky.social
Founder at @datalust.co, core maintainer for #serilog, original #autofac author, long-ago-.NET PM. Deep down the #rustlang rabbit hole. Also at https://nblumhardt.com/@nblumhardt.
Hope you're all doing okay out there ๐
29.07.2025 22:56 โ ๐ 2 ๐ 0 ๐ฌ 0 ๐ 0I can't believe it's so rare to get a terse, sensible, un-pushy issue raised ahead of a PR that feel the need to micro-blog when it happens, but here you go ๐ github.com/serilog/seri...
29.07.2025 22:56 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0Meanwhile, if you do want Serilog -> OpenTelemetry SDK bridging, watch or star github.com/serilog/seri...
31.05.2025 01:15 โ ๐ 2 ๐ 0 ๐ฌ 0 ๐ 0Building the Serilog sink for OpenTelemetry on OTLP, rather than the SDK, is turning out pretty well. It's super stable and in wide use at v4.2.0, while there's still no final SDK API for non-Microsoft loggers.
Dependencies aren't inherently bad, but **schedule dependencies** almost always are.
Work Sharing in Seq Clusters โจ
blog.datalust.co/work-sharing...
Just spiked out `Ctrl Space` `N` as a shortcut for this, feels pretty good; change is isolated enough we could get it into a patch.
Would it 100% nail this scenario for you?
Not quite the same thing (I'm talking about doing it without explicitly adding MCP tools), but thanks for the link ๐
26.05.2025 23:13 โ ๐ 0 ๐ 0 ๐ฌ 1 ๐ 0Probably the better default ๐ - for the app Iโm looking at, selectively enabling individual commands would make the experience safer, but access permissions behind the scenes would also cover it
26.05.2025 07:53 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0Would take some gymnastics but seems doable; I got this going for seqcli as a spike, it means access to the appโs features are fine-grained, vs all-or-nothing if itโs exposed as a regular command-line app
26.05.2025 07:35 โ ๐ 0 ๐ 0 ๐ฌ 1 ๐ 0No, to directly expose and invoke commands as MCP tools, using descriptions etc. from command metadata
26.05.2025 07:31 โ ๐ 0 ๐ 0 ๐ฌ 3 ๐ 0NuGet download stats for SeilogTracing showing 1.0M total
The first 1.0M is the hardest ๐
www.nuget.org/packages/Ser...
Has anyone investigated writing an MCP server binding for System.CommandLine in #dotnet?
Just poking around to keep up with what's going on in that space, seems like it could be an interesting quick bootstrapping option for existing codebases..?
Could be the wrong hemisphere for you ๐จ๐ฆ
23.05.2025 09:53 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0@fastmail.com, have been a customer for 10+ years via @datalust.co and have had no regrets
23.05.2025 09:51 โ ๐ 2 ๐ 0 ๐ฌ 1 ๐ 0Just moved from Gmail after just a crazy number of years. I still use and appreciate a bunch of Google products, but I'm wary of how centralized all these critical apps and services have become. Adopting alternatives from smaller, focused companies wherever I can at the moment...
23.05.2025 08:30 โ ๐ 1 ๐ 0 ๐ฌ 1 ๐ 0Happy Firefox user, here, but I need to spend more dev time in Chromium-based browsers. Is the Chromium project itself the go-to privacy-conscious option? Any others I should take a look at? ๐
23.05.2025 00:38 โ ๐ 1 ๐ 0 ๐ฌ 1 ๐ 0A command line interface
Some c# code
Weโre about to take C# to the next level!
#dotnet #csharp
Is hash-bang support possible? Kinda cool to launch directly as scripts on Linux
22.05.2025 04:06 โ ๐ 5 ๐ 0 ๐ฌ 2 ๐ 0After me spending years maintaining a .NET state machine library, we implemented the core node state machine in plain old C# ๐
Couldn't be happier - states as classes, trigger handlers as switch statements returning new states, it's such nice, clear, direct code I don't think I'll look back....
A screenshot of an `explain select distinct(Keys(@Properties)) from stream` query in the Seq UI. Indentation shows the "receive" operations outermost, and "send" operations indented below them. Receive operations are tagged with their merge strategy, e.g. "unordered", and the send operations with their partitioning strategy, e.g. "unconstrained".
Since data movement has such a huge impact on performance in distributed systems, we did a bit of work cleaning up Seq's query plans in 2025.1, and now explicitly show the send/receive operations that connect computations locally and over the network.
Still not perfect, but getting closer! ๐
Today's the day! Seq 2025.1 is ready at datalust.co/download, and on Docker Hub/ECR ๐
Find out what's new in the release announcement ๐ blog.datalust.co/announcing-s...
Congrats! 1M speaks of a lot of hard work ๐ ๐
07.05.2025 03:58 โ ๐ 1 ๐ 0 ๐ฌ 1 ๐ 0I just went through this, too, via the browser. Found the experience really enjoyable!
There's a lot of thought-leadering about WebSockets being obsoleted by this or that new HTTP feature - which, I think, means WebSockets are finally mature enough to build on ๐
Hello, world! ๐
24.04.2025 22:35 โ ๐ 4 ๐ 2 ๐ฌ 0 ๐ 0What I've been up to for the past year: blog.datalust.co/seq-clusteri..., and one of the toughest things to ship that I've ever worked on ๐
24.04.2025 22:08 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0Today's "quick-and-dirty" hack turned out quite nicely; a Seq relay (replication/forwarder) plug-in that properly handles traces, OTLP envelope properties, etc.: github.com/nblumhardt/s...
23.04.2025 06:35 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0Bizarrely, Windows Firewall doesn't like its forwarded requests: just lets them hang. Always some esoteric thing to go and chase..... ๐คทโโ๏ธ
17.04.2025 03:11 โ ๐ 0 ๐ 0 ๐ฌ 0 ๐ 0Most of the code is figuring out where to send a request, and detecting bounce loops; it's otherwise just: learn.microsoft.com/en-us/aspnet...
17.04.2025 01:04 โ ๐ 2 ๐ 0 ๐ฌ 0 ๐ 0Whoah, just spiked replacing Seq's internal cluster HTTP proxy with #aspnetcore's YARP, and once I remembered to register its service dependencies, it Just Worksโข๏ธ first try, WebSockets and all.
That's the mark of a nice piece of software, going to have to dig into this thing ๐
Thanks Sean! These are all really tough questions, BTW. I don't mean to suggest anyone's a bad person for making different choices, just that as things are today, we haven't collectively solved these problems in a way that feels constructive and fair to all involved. We'll get there, though ๐
16.04.2025 23:05 โ ๐ 1 ๐ 0 ๐ฌ 0 ๐ 0