Tara Oduya

Open ended with a floor rather than a ceiling worked for me. Rule is ten minutes minimum, no maximum, and I am not allowed to check anything during the ten. About seventy percent of the time the ten minutes turns into an hour, which is the whole point, and the other thirty percent I genuinely was not able to work and I have learned something about the day rather than sitting there resenting a timer.

The warm up cost is the thing that makes short blocks so bad for writing. You pay it every single time you restart.

Time is the cost people forget. Doing this properly by hand runs about four minutes a contact once you have a routine, so sixty is four hours of an evening you will not get back. A month of a finder tool costs less than that at any sane valuation of your time, and most of them have a small free allowance that covers a list this size.

One more path since people always suggest it: connecting on LinkedIn instead does not scale either, the weekly invitation cap for a normal account sits around a hundred and the acceptance rate on cold invites to people who barely use the platform is poor. For sixty offline manufacturers I would not bother.

Explain instead of produce is a clean rule, I can actually follow that. Do you let it write tests, or is that cheating too?

I would only add that most of us are not being bureaucratic on purpose. If a child writes an answer with no visible reasoning I genuinely cannot tell whether they understood it or copied it, and at that age half of them copied it.

I am sceptical of flat spotting as a ten minute explanation on a modern tyre. In my experience that resolves inside a kilometre or two unless the car has been parked for weeks. A shake that persists for ten minutes at a specific speed and then vanishes reads much more like ice to me, or possibly a wheel weight that fell off and a coincidence about the timing. Cheap test: drive it out of a heated garage one morning, or hose the wheels off before you leave. If it is still there on a warm wheel, get it rebalanced and check for a bent rim from a pothole.

Look outside the university entirely, because the department is a workplace and you are trying to make it be a community as well, which is a lot to ask of eleven people. My best two friendships from that period came from a weeknight football team and from volunteering at a repair cafe, neither of which contained a single academic. It also meant that when the department got politically unpleasant in my second year, which it will, I had somewhere to go that was not full of the same argument.

50k entries is small enough that you could also just clone the map into each worker and rebuild it periodically, if staleness of a few seconds is acceptable for a rate limit. Not elegant, zero contention.

Annihilation is a good call and I think we are describing the same shape from two sides. The narrator there is not kind though, which is why I did not reach for it.

Stop counting hours and count units of output, because hours are an input and you are optimising the wrong side of the equation. I switched to a rule of one meaningful thing shipped per day, defined the night before, and my hours did not change at all while my output roughly doubled. Most of what the extra hours were buying was polish on things that did not need it.

It's a u64 and an enum tag, so yes. Both approaches work here, this one is nicer when the condition gets complicated.

That's exactly it, and now the rule makes sense: keep a value not a reference. Compiles and the clone is gone.

Both can be true. Be upfront, but bias towards places with a half-life. I now write the long version as a post somewhere indexable and drop a link into chat, so the chat conversation happens and the artefact still exists in six months.

Keep it. Delete the fake features. Look again in six months.

Not hired yet so take this as a data point rather than advice, but the pattern I see in every one of these threads is that the people who got there had a non-code job that made them useful. Warehouse, retail, nursing, accounts. If you are managing a retail floor you already know rotas, stock, waste and shrink, and internal tools for those things are written by people who have never stood on that floor.

Twelve seconds and a decision is a good way to put it. Every soft measure I have tried became invisible in about ten days.

For the general debugging technique: comment out the body of the async method down to the smallest thing that still spawns, then add statements back until it breaks. Ugly, takes four minutes, and it beats reading the error when the error is 300 lines long.

The rewrite feels clean because you have forgotten the eleven decisions the old code encodes. Every one of them was a bug report or an edge case, and none of them are written down anywhere.

What actually happens: you rebuild 60% quickly, feel great, then reach the same boring part that made you stop seven months ago - the import parser, the billing edge cases, whichever it was - and abandon it again, except now you have two half-projects. Upgrade, ship one small visible thing, then decide with a working app in front of you.

Started this in February. What changed my numbers was tracking which specific questions the clicks came from - added a source parameter to the link in my profile and then noticed that essentially all my traffic came from two threads about one narrow problem, and none from the general 'what tools do you use' threads I was spending most evenings on.

Stopped answering the general ones. Same hours, and clicks roughly tripled over the next month.

The product photos are selling a small object. A cardioid mic already rejects most of what is behind it, so the filter is treating the direction that was least of your problem.

Write down what happens if the migration goes wrong halfway. Who reverses what, who tells the customers, who eats the refunds. Ten minutes on that beats any checklist.

Honestly, six weeks in I'd say clone freely and move on. Clone everything that isn't in a measured hot path, ship the thing, and go back and remove the clones later when you understand the code better. I lost about two months to fighting for zero-copy designs in code that ran twice a day.