Daniel Losch

825 posts

Daniel Losch

Daniel Losch

@cloudsider

... occasional head in the clouds. Software (everything .NET). Not an F5 type of person apparently.

Germany انضم Eylül 2009
300 يتبع34 المتابعون
Daniel Losch
Daniel Losch@cloudsider·
@Alan_Couzens While I would agree that optical hrv data is trash compared to (spot) strap measurements, recent data with the elevate v5 sensor hrv data in my case offers fairly good correlation to other health metrics (for example blood pressure). Plus advantage of not being spot check.
English
0
0
0
69
Alan Couzens
Alan Couzens@Alan_Couzens·
Just pulled my last 90 days of HRV data from my Garmin (wrist) data vs my 2 min morning HRV test with a heart rate strap. Not even directionally accurate... r = -0.13 ! Stop being lazy. Wear the damn strap.
Alan Couzens tweet media
Alan Couzens@Alan_Couzens

Anyone who has tried to use wrist heart rate on a run knows that it's basically a random number generator. This makes it completely worthless for #HRV, because even a few erroneous beats can have a HUGE effect on the HRV calculation. Manufacturers know this, so they apply HEAVY filters to their algorithms that can push things in the opposite direction - removing true variability... This means wrist HRV can read way above or way below true HRV in unpredictable ways! Bottom line: If you care about HRV accuracy (you should!), the only acceptable method is a morning test with a heart rate strap.

English
20
10
182
40.4K
Mads Kristensen
Mads Kristensen@mkristensen·
I'm really enjoying GPT-5.2-Codex as a supplement to Claude Opus 4.5
Mads Kristensen tweet media
English
20
5
154
17.2K
Daniel Losch
Daniel Losch@cloudsider·
@hjluks I was always sceptical about wearables generating hrv data. But (I own a @Garmin watch with Elevate v5 sensor) the correlation with other (medical/established) health metrics absolutely positively surprised me. I trust the data the watch collects. Valuable for training & health.
English
0
0
0
1.7K
Howard Luks MD
Howard Luks MD@hjluks·
Heart rate variability (HRV) has moved from an elite athlete tool to a mainstream health metric. Should it? Your smartwatch tracks it. Your Ultrahuman or Oura ring measures it overnight. Apps and strap manufacturers claim they can predict everything from overtraining to illness. Some of that is useful. Some of it is noise. This post will explain what HRV measures, how to interpret it correctly, and when it’s helpful versus when you’re overthinking a number that doesn’t matter as much as you think. howardluksmd.substack.com/p/heart-rate-v…
English
10
18
197
35.1K
Daniel Losch
Daniel Losch@cloudsider·
@pierceboggan @code The default keyboard shortcuts esp with agent mode are … suboptimal would be a euphemism.
English
0
0
0
38
Pierce Boggan
Pierce Boggan@pierceboggan·
VS Code is the best UX for managing your coding agents! Give the latest features in @code a try in Insiders. What can we do to make it even better?
Pierce Boggan@pierceboggan

New in @code Insiders: dedicated agent management UX with "Chat: Open Agent Sessions Window". The UX for managing your agents is here in @code :)

English
10
4
32
3.8K
Adrian Franke
Adrian Franke@adrianbb89·
das ist ein Spiel, in dem die sinnvolle sportliche Analyse Mitte des 3. Viertels abbricht
Deutsch
40
1
436
27.1K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
Supporting windows is just impossible. Does windows have any native sandboxing technology. I'm sure they do.
English
8
0
7
5K
Daniel Losch
Daniel Losch@cloudsider·
I enjoy working with coding agents, and they speed up several of my tasks (some of which I totally enjoyed doing, and some of which I never enjoyed having to do) tremendously, I consider them very helpful. But boy, they have zero understanding, and they don't even know so often.
English
0
0
0
34
Daniel Losch
Daniel Losch@cloudsider·
@mkristensen Were that completely non-obvious issues which are really hard to get right? I suppose you get the obvious and not so obvious stuff quite right by yourself.
English
0
0
0
145
Mads Kristensen
Mads Kristensen@mkristensen·
I ran the Visual Studio profiler agent on my code and optimized the performance with these results: - Large documents now parse 5x faster - Memory allocations reduced by 94% - Validation complexity changed from O(n²) to O(n)
English
9
0
98
10.5K
Daniel Losch
Daniel Losch@cloudsider·
I might be wrong but there is a huge difference between AI being smart enough to displace lots of jobs (99% of jobs are basic 90+% of the time) and AI becoming way smarter than humans. I see the first one arriving but ... yes, AI is quick and has more and more context, but that doesn't fix the fact ... it often seems to lack understanding. It gets better and better at tasks where knowledge leads to outcomes. Knowledge and understanding are two different things. Coding agents have very quickly gotten way better at the knowledge part. For me they still seem to be lacking or stagnating in the understanding part.
English
0
0
0
73
Dr. Mike Israetel
Dr. Mike Israetel@misraetel·
When (likely in 2026 or 2027) AI supersedes all humans in logical intellect, a very interesting question to ask the AI maye be this: "With the increasing ubiquity of computer-based AI agents for office tasks and the impending arrival of robotics for physical tasks, how would you make humans alive today the most productively useful they can be to bettering our human/machine civilization?" We keep saying machines will be so smart that they'll displace us from our current jobs. Surely they'll be smart enough to find even better (better paying, safer, more rewarding, more fun) jobs for us than, in our not so impressive intellectual state, we did? Been reading a lot of @alexwg's X posts lately. Got me thinking.
English
34
1
147
14.6K
Daniel Losch
Daniel Losch@cloudsider·
Hey @GeminiApp can you pls add an option to turn a temporary chat into a normal one? I often start chats which I intended to dispose but turn out to be super interesting. A button to export a full chat to markdown would be cool but I guess that's being excluded intentionally?
English
0
0
0
39
Daniel Losch
Daniel Losch@cloudsider·
@DamianEdwards @kzu Mine has support for this while (hopefully) being fairly defensive about what to delete (it should validate the tfm).
English
0
0
1
19
Damian Edwards
Damian Edwards@DamianEdwards·
@cloudsider @kzu no it doesn't, it only looks for output paths for TFMs the project currently uses, but that's a good feature request! It's *slightly* complicated as it means purge will need to guess if a discovered output path has other TFM variants. Logged: github.com/DamianEdwards/…
English
1
0
1
45
Damian Edwards
Damian Edwards@DamianEdwards·
Pushed a new release of dotnet-purge! Introduces parallelism for faster purging, along with support for a `--dry-run` flag & proper handling of shared projects. nuget.org/packages/dotne…
English
5
15
73
8.9K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@davidfowl Wow, that was very uneventful. I ran the docker container. It worked, I now have a OTEL viewer. That was so easy it's almost not fun.
English
1
1
14
7.4K
Darren Shepherd
Darren Shepherd@ibuildthecloud·
@davidfowl I'm trying aspire today because I want your otel viewer. Wish me luck. Or actually I hope this goes well for you :)
English
1
0
7
1.5K
Daniel Losch
Daniel Losch@cloudsider·
@DamianEdwards The tool does full msbuild property evaluation and (depending the targets are in the vs install) als cleans framework style. I was never happy with dotnet clean
English
0
0
0
34
Daniel Losch
Daniel Losch@cloudsider·
@DamianEdwards @kzu Does your tool clean old tfms after you upgrade? So I upgrade my project to net10.0 and all the old net8.0 and net9.0 won’t get cleaned (at least not by dotnet clean)
English
1
0
0
27
Damian Edwards
Damian Edwards@DamianEdwards·
@kzu I frequently bounce between projects/repos using artifacts output path or not or custom configuration names, but if you have something that works for you, all power to you 😁
English
2
0
1
157
Daniel Losch
Daniel Losch@cloudsider·
Most stuff from the dotnet/aspnetcore/runtime/aspire repos have dedicated artifacts directories which are not below bin. Maybe because they know, once you have the build output it’s nontrivial to get rid of it. Or after you upgrade the tfm, good luck cleaning the obsolete folders.
English
0
0
0
25
Daniel Cazzulino 🇦🇷🗽
@DamianEdwards Sure, I get that. I just haven't come across many projects that customize builds to the point of not even having bin/obj somewhere under the repo root
English
3
0
1
180
Daniel Losch
Daniel Losch@cloudsider·
@unusual_whales Maybe Taylor Swift actually wrote Opalite as an ode to vibe coding … „eating out of the trash“
English
0
0
0
179
unusual_whales
unusual_whales@unusual_whales·
Cursor CEO has said that vibe coding builds ‘shaky foundations’ and eventually ‘things start to crumble,' per FORTUNE
English
318
288
6.5K
804.8K
Daniel Losch
Daniel Losch@cloudsider·
@davidfowl Opus 4.5 is truly good. For smaller tasks, Gemini 3 flash (preview) in copilot is also very great bang for buck. What is your experience with opus when working on an existing, complex codebase? For me it is more like hit and miss, even for very specific tasks/prompts.
English
0
0
0
369
Daniel Losch
Daniel Losch@cloudsider·
@mkristensen Like it. Any chance it can write relative paths (well I guess to the sln(x)) to the favorites.json?
English
0
0
0
20