David Strauß

@mitsuhiko @steipete Multiple checkouts reminds me of Fossil where this is the normal workflow: fossil-scm.org/home/doc/tip/w…

I talked with @steipete yesterday and we both realized that we gave up on worktrees and just use multiple checkouts. Turns out, we're simple people. (And yes, I had a vibecoded worktree manager. Ended up just not using it)

@nateberkopec @Alan_Couzens I was also very confused. I think you need to subscribe (see screenshot) in order to get access to all threads.

@Alan_Couzens I must be missing something. Signing up for that Discourse only allows me to see the "Admin" forum, I can't view that thread.

*Every* 4-6 week training block should have a benchmark that you're working toward... After all, big goals are nothing more than tiny benchmarks stacked over time. Stack the small wins.

@Alan_Couzens What HRV measurement protocol should we use? In the morning, with a chest strap I would think. Do you have a no fuss software recommendation?

Interesting discussion going on the #MADcrew forum right now around what... - a healthy nervous system - a "broken" nervous system ..each look like. Trust me, not something you want to mess around with! All athletes should track and understand #HRV ! forum.madcrew.app/t/what-does-a-…

@johngetstrong @cary_carytb @inaki_delaparra @Alan_Couzens So if you go for a walk, you explicitly start an activity on your watch?

@cary_carytb @inaki_delaparra @Alan_Couzens Just the measurable activities - so NEAT not included. I imagine NEAT probably adjusts up and down based on physical activity and caloric intake, but I imagine that's probably a very small factor in terms of fitness.

I ran my first marathon today in a time of 2hr 58min! 😀 My longest run prior to this was 24km, a couple of weeks ago. Absolutely thrilled with this result, especially given how little I knew about how well I would tolerate a race this long. Found that the limiter was more muscular than aerobic. Heart rate averaged 167, which is around 89% of my max heart rate. Felt aerobically controlled pretty much throughout the entire race. Heart rate dipped on the downhill portions in the final few kms as my calf became the limiter for how hard I could push — the mind and aerobic system were ready, but the muscles said no to anything more. Given my 1:24 HM in May, I knew this kind of time was possible, but it still surprised me that the durability was there for a sub-3hr Mara. Some more details from the Mara: * No breakfast beforehand. Sometimes meals + exercise can cause a rapid crash for me, and I didn’t want to risk it. Took a gel 15 minutes before race start instead. * Woke up as late as possible (5:35 for a 6:31 start). Ended up arriving at the start line at 6:30, and crossed around 6:35. Figured the extra sleep was more important than trying to push to the front of my start group. * 7 gels (25g CHO) taken during the Mara — approximately every 5km. * Carried a small 250ml water flask. This got me through the first 1hr 20min without needing to pick up water. I’d consider carrying even more next time. * Running volume has been 40km/week over the past 3 months. This is down considerably from an average of 70km/week over the preceding year. * Average kcal burned per day this year is around 1,500 kcal — roughly 40% running (8km/day), 35% walking (about 10km/day), and the remainder from gym workouts and conditioning (i.e. Hyrox sessions). * Feel good energetically post-race. Muscles are sore, and I definitely hobbled home. But surprised at the lack of "depleted" feeling post race. Focus for the rest of the year, will be gradually increasing daily energy output with walking, running, and some indoor cycling (I just got a Zwift Ride).

@_m_b_j_ What are the fundamentals in your experience? Has it more to do with project specific aspects like correct data schema, or the technical side like a fast test suite?

Your (already longer running) software project is slowing down, and you want to speed up? You want to spend more on it to speed it up again? Make sure you fix fundamentals with the spending FIRST. Thank me later.

@triskweline Der Link funktioniert bei mir leider nicht.

BTW our offices are 100 meters apart :) Read Tuxedo's case study here: mytuxedo.de/index.php/s/ir…

If you're eying Linux for @Rails work: It's not just @dhh. At @makandra_de we've been partnering with @TUXEDOComputers to supply Ubuntu-certified laptops to our Rails team. Running the same OS in dev+prod eliminates failure modes & reduces the pressure to Dockerize everything.

@basiszwo I think some clients pre-fetch links. So if your GET HTTP handler makes a state change, there is a problem. I do this: GET /login/:token renders a page +form. If JS is enabled, auto-submit to POST /login which does the state change. If not, the user submits the form manually.

I am facing a weird behaviour and am basically clueless. I have a passwordless sign-in known from slack and others. works fine except for windows and outlook. it seems that when clicking the link in outlook, the request is done twice which renders the sign-in request useless, as I burn the token after first usage. I could do some workarounds but I‘d like to know why outlook does this and this behaviour is fixable. anyone? #Windows #Outlook

@wolax @Ubiquiti Was für ein Rack ist das? Bei uns wird es langsam ernst 😅

Did a whole on-site installation of @Ubiquiti UniFi gear without even taking my laptop out of my bag. Remarkable software experience. ✨

@_m_b_j_ Thanks, that sounds better than what I do right now. Do you keep two versions for every file, so you can migrate "down" or are these migrations all irreversible?

@stravid We have a `db/{views,functions,whatever}` directory with files that create these objects. Migrations execute that files - and have diffs. Also make sure you maintain a `db/structure.sql` dump for booting local DBs and see "evaluated" diff.

@_m_b_j_ I manage database schema changes via migrations. Changes to views / procedures are a pain to review since there is no diff. It must be created manually outside the PR. How do you handle this in your team? I would appreciate any pointers towards a less painful approach.

@kerckhove_ts As someone super jealous of Sydtest I approve the last point 😂

My colleague making a TODO list for a Rust project:

@Hetzner_Online wann schätzt ihr, wird Debian 12 bei euch in der Cloud verfügbar sein?

@ederbit I have used OBS successfully in the past.

Everyone seems to be using Loom to record professional screencasts. Not supported in Linux. Any recommendations?

@GergelyOrosz @realGeorgeHotz This should have a huge disclaimer. Software engineering is not done like that. An intern trying to make it look like it works is cool, we all play and learn. But if this is the acceptable level of what we call engineering, no wonder our industry goes to shit.

I do recommend to follow along @realGeorgeHotz who is being one of the most transparent software engineers at Twitter as of now, and clearly someone motivated & determined to move blazing fast. We can learn a lot on what the reality of working with a large codebase is from him.

@_m_b_j_ @cercerilla Non-Haskeller question: What language attributes are responsible for that?

@cercerilla I've experienced this to but attribute it to the fact that haskell forces you more to deal with the problem space of your domain early and more complete than with languages you can cut corners more easily.

@wolax Danke für die netten Worte!

@stravid Klasse Beitrag voller interessanter Aspekte. Herzlichen Dank für's ausführliche Niederschreiben.

Update to my „Building Besserliste“ tweets from earlier this year: My digital shopping list is complete and in use 🎉 I wrote a report about the process and things I learned, unfortunately currently only in German. stravid.com/60/bessere-sof…

@davetron5000 @brandur In my opinion, every background job needs to be written to the DB. Why have a background job in the first place if it's okay to lose it? Never understood the blind Sidekiq ideology in the Rails world. It WILL cause data inconsistency if you use it as advertised without extra work

@davetron5000 If something needs to be transactional, use a transaction. No room for Redis or any other foreign datastore apart from your DB. Most apps will be fine with DB based background jobs. You are an outlier? Use a transactional job drain. @brandur wrote about it brandur.org/job-drain

@tdvorak Wir waren in die andere Richtung unterwegs. Gastein war ordentlich steil am Ende und wir haben den Zug um 5 Minuten verpasst :D Dafür war die Abfahrt in der Abendsonne nach Obervellach eine schöne Belohnung.