Sign in

Daniel Marbach

@daniel.openplace.net
466 followers 416 following 2.7K posts
PostsRepliesMedia
Daniel Marbach @daniel.openplace.net · 30/09/2026
My vacation experiment... Makes your AI agent think like the architect everyone rolls their eyes at, until they're right. danielmarbach.github.io/turtleneck/ #AgentSkills #aiagent
110
Reposted by Daniel Marbach
Kevin Gosse @kevingosse.net · 29/09/2026
I published a new article: the hidden trap of fixed buffers in C#. It explains a quirk of fixed buffers that could easily bite you, if you're one of the few people who still have to use that feature. minidump.net/the-hidden-t...
minidump.net
The hidden trap of fixed buffers in C#
After all, if they’re marked as unsafe, there’s gotta be a reason, right?
285
Daniel Marbach @daniel.openplace.net · 21/09/2026
Azure Service Bus cannot autoforward a session-enabled subscription. I explored a transactional bridge that keeps one session-enabled endpoint queue and a recoverability model. www.planetgeek.ch/2026/09/21/a... #azure #dotnet
planetgeek.ch
Azure Service Bus sessions: Pub/sub without autoforwarding
TL;DR: Azure Service Bus does not support autoforwarding from a session-enabled subscription. A direct workaround is to process each subscription as an application input, but that spreads session lock...
021
Daniel Marbach @daniel.openplace.net · 15/09/2026
Delayed retries can break Azure Service Bus session ordering. If A1 fails, A2 and A3 may overtake it. Strict order may require a hold-back. www.planetgeek.ch/2026/09/15/a... #azure #dotnet
planetgeek.ch
Azure Service Bus sessions: Delayed retries may need a hold-back
TL;DR: Most workflows are better off accepting out-of-order delivery and modeling the business process to cope with it. If strict order is a real requirement, however, Azure Service Bus sessions are o...
022
Daniel Marbach @daniel.openplace.net · 02/09/2026
Topology migrations are safest when they are boring. While helping a customer migrate an Azure Service Bus topology, we moved one event stream at a time: prepare the new path, switch publishers, observe, then remove the old path. www.planetgeek.ch/2026/09/02/a... #azure #dotnet
planetgeek.ch
Azure Service Bus: Migrate topology one event at a time
TL;DR: Azure Service Bus can run an old and a new delivery path side by side while you migrate one event stream at a time. A delivery path is the route from a publisher, through a topic and subscripti...
000
Daniel Marbach @daniel.openplace.net · 25/08/2026
One backed-up event stream on a shared Service Bus topic can send unrelated API latency through the roof. I've seen it happen. A shared topic concentrates risk. Design topics as failure boundaries. www.planetgeek.ch/2026/08/25/a... #azure #dotnet
planetgeek.ch
Azure Service Bus: Designing for blast radius
TL;DR: Blast radius means how much unrelated traffic is affected when one messaging resource reaches a limit or stops making progress. If every event type shares one Azure Service Bus topic, a backed-...
010
Reposted by Daniel Marbach
Tanya Janca | SheHacksPurple @shehackspurple.bsky.social · 22/08/2026
0123
Daniel Marbach @daniel.openplace.net · 14/08/2026
One topic, many subscriptions in Azure Service Bus. A shared topic can be a good trade-off. But filters, fan-out, and storage work accumulate there, and topic depth shows when delivery falls behind. www.planetgeek.ch/2026/08/14/a... #AzureServiceBus #NServiceBus
planetgeek.ch
Azure Service Bus: One topic, many subscriptions
TL;DR: A single Azure Service Bus topic can be a reasonable starting point for one-to-many messaging. The topic receives published messages, and subscriptions provide separate delivery views for consu...
022
Daniel Marbach @daniel.openplace.net · 03/08/2026
I’ve been using Azure Service Bus since the Service Bus for Windows Server days. My new series covers Azure Service Bus lessons learned the hard way. First up: filter count. At scale, the number of rules can matter more than the filter type. www.planetgeek.ch/2026/08/02/a... #azure
planetgeek.ch
Azure Service Bus: Count your filters
TL;DR: Azure Service Bus can publish one message to a topic and deliver it to multiple consumers. A topic is the shared destination, a subscription is a consumer-specific view, and a filter decides wh...
011
Reposted by Daniel Marbach
Patricia Aas 🐢🏳️‍🌈 @patigallardo.bsky.social · 26/07/2026
1/ The Chinese Open Source model Kimi-k3 from Moonshot AI is claiming it will release it’s weights on Monday. «The full model weights will be released by July 27, 2026.» What does this mean in normal speak and what are the possible implications on stocks and politics? www.kimi.com/blog/kimi-k3
kimi.com
Kimi K3 Tech Blog: Open Frontier Intelligence
Kimi K3 is the world's first open 3T-class model — frontier performance across coding, knowledge work, and reasoning, with native multimodality and 1M context.
13512
Daniel Marbach @daniel.openplace.net · 21/07/2026
Finally something that makes the open router added markup worth it openrouter.ai/docs/guides/...
openrouter.ai
Auto Router - Intelligent Model Selection
Automatically select the best AI model for your prompts using OpenRouter's Auto Router powered by task type rankings.
000
Daniel Marbach @daniel.openplace.net · 20/07/2026
How do you keep performance gains without fooling yourself? The final post in my performance and benchmarking series covers useful regression gates, benchmark variance, noisy CI runners, and why human judgment still matters. www.planetgeek.ch/2026/07/20/k... #dotnet #performance
012
Reposted by Daniel Marbach
ZurichJS @zurichjs.com · 08/07/2026
So happy to have @debbie.codes joining us this September 💛 Platform Engineer doing Applied AI at Zephyr Cloud, ex-Microsoft Playwright wizard, and just genuinely one of the good ones in this community. Come say hi and catch her talk!
0116
Reposted by Daniel Marbach
Clemens Vasters @clemens.vasters.com · 06/07/2026
UEFA www.uefa.com/news-media/n...
uefa.com
UEFA statement on the Balogun case
Statement on decision related to Folarin Balogun
181
Daniel Marbach @daniel.openplace.net · 02/07/2026
I wrote about what happens when a message handler outlives the queue lease. Renewing sounds easy until you need a renewer, mutating receipts, coordinated delete, and shutdown juggling. The first loop is easy. The edges are where the work lives. www.planetgeek.ch/2026/06/29/w...
011
Daniel Marbach @daniel.openplace.net · 22/06/2026
A benchmark win is not the finish line. The real workload may tell a different story. Profile the larger execution path again to see whether the improvement compounds. www.planetgeek.ch/2026/06/22/a...
010
Daniel Marbach @daniel.openplace.net · 20/06/2026
The first version of a message pump is easy. The tenth version contains: retries idempotency transactions topology monitoring The hard part isn’t the loop. It’s owning every failure mode behind the loop. www.planetgeek.ch/2026/06/20/m... #dotnet #messaging
010
Daniel Marbach @daniel.openplace.net · 15/06/2026
The biggest red bar isn’t always the biggest problem. Profilers reveal costs. We set priorities. www.planetgeek.ch/2026/06/08/r... #dotnet #performance
011
Daniel Marbach @daniel.openplace.net · 13/06/2026
We inherited a legacy medical invoicing system. The business wanted to expand into web shops. Same invoices. Very different failure modes. Here’s what a message queue taught us about temporal decoupling 🧵👇 www.planetgeek.ch/2026/06/13/l...
142
Daniel Marbach @daniel.openplace.net · 10/06/2026
Production code is too noisy. Benchmarks are too focused. Between them sits the profiling harness. Small enough to isolate a code path. Realistic enough to expose allocations and CPU costs. Visibility first. Comparison second. www.planetgeek.ch/2026/06/01/b... #dotnet #performance
010
Daniel Marbach @daniel.openplace.net · 08/06/2026
The first benchmark I wrote looked deceptively easy. A class. A few attributes. A method. The hard part isn't writing benchmarks. It's knowing what deserves one. Profiles identify. Benchmarks validate. Production teaches. www.planetgeek.ch/2026/05/27/s... #dotnet #performance
020
Daniel Marbach @daniel.openplace.net · 18/05/2026
Openclaw Patch has now #dotnet support thanks to a collaboration between me github.com/openclaw/cla... and Simon github.com/openclaw/cla...
github.com
010
Daniel Marbach @daniel.openplace.net · 17/05/2026
😂
010
Reposted by Daniel Marbach
simoncropp.bsky.social @simoncropp.bsky.social · 16/05/2026
Proposal: Open Source Maintenance Fee for Verify github.com/orgs/VerifyT... Any feedback is appreciated #dotnet
github.com
Proposal: Open Source Maintenance Fee for Verify · VerifyTests · Discussion #1731
The Verify project is considering adopting the Open Source Maintenance Fee (OSMF), and we'd like to open it up for discussion before committing to anything. Verify has been freely available since 2...
4167
Daniel Marbach @daniel.openplace.net · 04/05/2026
Preparing for a CfP? Read this first einarwh.no/blog/2026/04...
einarwh.no
What to say
010
Reposted by Daniel Marbach
Particular Software @particular.net · 03/05/2026
It's not too late to register for Poornima's workshop - Beyond Publish/Subscribe: Messaging Patterns in Practice
techorama.be
031
Reposted by Daniel Marbach
Barry Dorrans @blowdart.me · 21/04/2026
.NET has an out of band update today to fix CVE-2026-40372, an Elevation of Privilege, which, in some cases, could allow an attacker to forge authentication tickets, or decode authentication tickets or other protected data. github.com/dotnet/annou...
github.com
Microsoft Security Advisory CVE-2026-40372 – ASP.NET Core Elevation of Privilege · Issue #395 · dotnet/announcements
Microsoft Security Advisory CVE-2026-40372 – ASP.NET Core Elevation of Privilege Executive Summary: A bug in Microsoft.AspNetCore.DataProtection 10.0.0-10.0.6 NuGet packages can give an attacker th...
22417
Reposted by Daniel Marbach
Todd H. Gardner @toddhgardner.com · 16/04/2026
Most engineers undervalue their time. You have more important work to do than re-inventing the wheel.
063
Reposted by Daniel Marbach
Weekday Jokes @weekdayjokes.bsky.social · 06/04/2026
Mortal Kombat was actually based on a Scandinavian church song. It was a Finnish hymn
12664180
Reposted by Daniel Marbach
KanDDDinsky @kandddinsky.bsky.social · 17/03/2026
We don't need another DDD intro talk. We need your stories. The refactoring that almost sank the project. The EventStorming that exposed real org structure. The model that collapsed under load. KanDDDinsky 2026 is built on experience reports. CFP closes March 21: sessionize.com/kandddinsky-...
033
Daniel Marbach @daniel.openplace.net · 23/02/2026
I can ship some. A metric ton in your living room enough?
020
Reposted by Daniel Marbach
Kirill Osenkov @kirillosenkov.com · 21/02/2026
PerfView can record all managed exceptions in all processes system-wide thanks to the .NET ETW provider, with callstacks. I built a viewer for these .etl files:
Tree view with the processes on the left, timeline of exceptions in processes up top, data grid with the exception details in the center and current exception call stack at the bottom
3398
Daniel Marbach @daniel.openplace.net · 21/02/2026
But at the end of the day you realize once you mastered those things that what counts is that you started somewhere and learned what to try, use and dismiss and how to separate BS from usefulness
010
Daniel Marbach @daniel.openplace.net · 21/02/2026
Honestly I believe it doesn't really matter where you start. It's similar to for example trying to become a perf expert or a gym rat. If you research those topics you realize there is so much that you just want to stick your head in the sand.
310
Daniel Marbach @daniel.openplace.net · 21/02/2026
This is a good summary that describes the strengths of both approaches youtu.be/41UDGsBEjoI?...
youtu.be
Codex 5.3 vs Opus 4.6: The Benchmark Nobody Expected. (How to STOP Picking the Wrong Agent)
YouTube video by AI News & Strategy Daily | Nate B Jones
010
Daniel Marbach @daniel.openplace.net · 21/02/2026
Ok. I would have recommended then to start with codex. I think it is superior to Claude but some others might disagree. The cool thing with openai is that you can use it in open code. With Claude that is against terms and might get you into banning
200
Daniel Marbach @daniel.openplace.net · 21/02/2026
Do you already have a chatgpt subscription?
100
Daniel Marbach @daniel.openplace.net · 21/02/2026
This is an excellent wakeup call m.youtube.com/watch?v=q6p-...
m.youtube.com
Going Slower Feels Safer, But Your Domain Expertise Won't Save You Anymore. Here's What Will.
YouTube video by AI News & Strategy Daily | Nate B Jones
000
Daniel Marbach @daniel.openplace.net · 21/02/2026
One thing that really helped me was to embrace the fact that I have years of experience, trust my instincts, know my and the tools limits and then experiment while keeping a sane scepticism. And then try and error. Rinse and repeat
010
Daniel Marbach @daniel.openplace.net · 21/02/2026
Try open code with something like open router or synthetic. There is so much more than just anthropic and openai
110
Reposted by Daniel Marbach
Andy Kelk @andykelk.net · 21/02/2026
I like @willvelida.com's definition of AI coding as HVE (hyper velocity engineering). Important point being velocity which implies *direction* not just speed. #DDDMelb
053
Reposted by Daniel Marbach
Kelsey Hightower @kelseyhightower.com · 10/02/2026
We are asking developers to do more with less, including less pay. Either we have managed to cut cost without reducing output, or we have increased output, but have very little to show for it. I'm sure there is another narrative around creating new opportunities, but I'm not hearing about those.
37711
Reposted by Daniel Marbach
qdot @buttplug.engineer · 04/02/2026
inspired by CLAUDE.md, I’ve started putting markdown files named after coworkers into work code repos so I can remind them to stop doing shit to the codebase that annoys me for some reason they’re all mad at me now, which means ill be adding commands to JEREMY.md for an attitude adjustment
6814176
Daniel Marbach @daniel.openplace.net · 03/02/2026
The enshitification of chatgpt medium.com/write-a-cata...
medium.com
OpenAI’s Latest Move Confirms My Worst Fear
A few days back, I saw this mail from OpenAI.
010
Daniel Marbach @daniel.openplace.net · 01/02/2026
I use alpha progression. Pretty neat app
100
Reposted by Daniel Marbach
Laurent Kempé 🚀 @laurentkempe.com · 01/02/2026
Vercel's AGENTS.md approach outperforms skills by embedding comprehensive Next.js docs into the agent context, avoiding decision points and ordering issues while ensuring consistent availability across turns — though skills still shine for focused, user-triggered tasks like upgrades or migrations.
vercel.com
AGENTS.md outperforms skills in our agent evals - Vercel
A compressed 8KB docs index in AGENTS.md achieved 100% on Next.js 16 API evals. Skills maxed at 79%. Here's what we learned and how to set it up.
031
Reposted by Daniel Marbach
Particular Software @particular.net · 29/01/2026
Join @jasontaylordev for a live webinar: Effortless Distributed Systems with Aspire, February 4 at 01:00 UTC. Learn how Aspire orchestrates services with reliability & observability, NServiceBus improves resilience, and Azure Developer CLI streamlines dev→cloud.
particular.net
Effortless Distributed Systems with Aspire
See how Aspire, NServiceBus, and Azure Developer CLI simplify building distributed systems in this interactive session. …
011
Daniel Marbach @daniel.openplace.net · 24/01/2026
Congratulations and thanks for everything you did!
010
Daniel Marbach @daniel.openplace.net · 24/01/2026
Be sure to ping this swiss dude. Happy to torture you and your husband 🤪
100