https://www.europesays.com/it/744782/ Milan, Amorim e tutti i dubbi da Moreira al turnover e i moduli #adattabilità #amorim #atalanta #bournemouth #Calcio #camarda #cambi #campionato #chukwueze #derby #difesa #diversi #domani #dubbi #DubbiAmorim #estupinan #Football #infortunio #infrasettimanale #IT #Italia #Italy #lecce #milan #modric #moreira #rabiot #rebus #rosa #rossoneri #salisburgo #san #SanSiro #serie #settimana #siro #Soccer #spinto #Sport #Sports #tecnico #titolare #trittico #turnover #udine
Search
7 results for “rebus”
-
"Le slide del decimo convegno ONIF a Genova"
Il 16 ottobre 2026 l’Osservatorio Nazionale per l’Informatica Forense ha tenuto il suo decimo convegno a Genova. L’associazione promuove costantemente la figura dell’informatico forense e l’applicazione di me… https://ift.tt/RHO83kY
-
ready for the Rebs thanks to @[email protected] 🤘
-
Is Diageo (LSE:DGE) Still An Income Name After Its Rebased Dividend Went Ex This Month?
Highlights Diageo (LSE:DGE) Consumer Defensive Diageo PLC (LSE:DGE) 1790.00 GBX +34.000 ↑ 1.936% Last Updated at: 2026-10-16T15:39:00Z traded…
#EuropeSays #Britain #Europe #EU #Diageo #diageo
https://www.europesays.com/britain/151705/ -
Sprzedaję :) / Polecam :) / Twarda oprawa, niebieska wstążeczka, kultowa seria "Salamandra"/ #Sprzedaż
....................................................................................
https://www.vinted.pl/items/10508569956-ptasiek-william-wharton-wyd-rebis-salamandra-1993-twarda-oprawa
..................................................................................... -
For two weeks, Claude planned and reviewed my code changes and Xiaomi's MiMo-V2.6-Pro wrote them. That was 95 work orders across 14 of my repos, and 62 of them merged. At list prices, MiMo's side of those 62 changes cost $14.22. The same tokens at Claude Opus 5.5 prices come to $355.
The catch is time. A merged change took 52 minutes from order to merge (median). MiMo was working for less than half of that.
The setup
A dispatcher I wrote (about 10,000 lines of stdlib Python) runs each order in its own git worktree and branch, inside a macOS seatbelt sandbox that can only write to that worktree. Claude writes the order, MiMo implements it and runs the project's checks, Claude reviews the diff and either sends it back with feedback or merges it locally. Nothing gets pushed without me.Every order leaves a state file and the worker's full event log, with timestamps and token counts for each model call. All the numbers below come from those files, for orders created 11 to 23 October 2026.
Where the time went
StageShare of wall-clockMedianWaiting in the queue5%0.1 minMiMo working49%30.7 minBetween rounds (review, feedback, waiting)27%6.4 minLast round finished → merged20%11.2 min
I could rebuild the full timeline for 33 merged orders (four more sat overnight and I left them out).
The orders that passed on the first round make it plain. MiMo worked for 16 minutes (median) and the order took 42 minutes end to end.Most of the extra 26 minutes is the orchestrator not looking. Its instructions said to check the queue with
sleep 595, then review, then merge. The long interval was there to save Claude tokens. It also left up to ten minutes of dead time every time a worker finished. The merge step adds more: it rebases onto whatever landed in the meantime and reruns the full test suite.None of this is the model's fault. It's plumbing.
Is the model itself slower?
Per model callMiMo-V2.6-ProClaude Opus 5.5Calls measured6,6404,537Seconds per call, median6.77.9Seconds per call, mean21.813.2Output speed47 tok/s81 tok/sContext per call, median153K467K
Less than I expected. I compared MiMo's turns with Claude Opus 5.5 turns from my own Claude Code transcripts. They're different tasks, so read this as a rough comparison, not a benchmark.
The typical call is as fast. The slow ones are slower, which pulls the mean up. A MiMo attempt ran 37 calls and 15 minutes (median).The context column matters later. MiMo starts every order from a fresh worktree, so it carries about a third of the context my long Claude sessions do.
The real gap: rework
Rounds to mergeOrdersMedian tokens1243.6M2208.9M3814.8M4 or more1031.0M
Only 24 of the 62 merged orders (39%) went through on the first round. The rest needed one to six more. Each round means another attempt, another review and another wait, and the token bill climbs fast:
I read the review feedback that sent orders back. Nearly all of it falls into five groups:- Failing silently: a missing column, a join that matched zero rows or a button that wasn't found, all treated as success.
- Data edge cases: duplicate rows, a date that's
objecton one side andstron the other, and tickers likeNAandNULLthat pandas quietly reads as NaN. - Tests that can't fail. One asserted that a static header string was present in the output. Another wrote into the repo's real output folder. One took 113 seconds against a 60-second limit.
- New behaviour with no test, or a runbook left describing the old behaviour.
- Matching too loosely: a substring check for
selectedthat also hitsunselected, or "click the first OK button on the page".
None of these are hard to get right. They weren't in the order, and the worker didn't go looking for them. A stronger model goes looking more often. That's most of what you pay Claude for.
The cost math
Here's the token mix across the 62 merged orders: 97.8% cache reads, 0.56% output. An agent re-reads its whole context on every call, so the token count is roughly calls × context size. It has little to do with how much code gets written.That makes the cache-read price the one to compare, and the headline price misleads:
Per 1M tokensMiMo-V2.6-ProClaude Opus 5.5RatioInput$0.435$4.009×Cache read$0.0036$0.2056×Output$0.87$20.0023×
On my mix the blended ratio is 25×. A merged change cost MiMo $0.17 at list price (median) against $3.83 for the same tokens at Opus prices. Even the orders that took four or more rounds stayed far cheaper than handing the job to Claude. And Claude doing the work inside my planning session would carry three times the context per call, so its real bill would be higher still.I pay for both through subscriptions, not per token, so list prices are a yardstick here, not my invoice. Xiaomi's Token Plan page doesn't say how cache hits count against the monthly quota. Check that before you rely on a ratio.
The money that actually leaks is on the orchestrator side. Every time Claude wakes up to poll, it re-reads its own large context. Ten polls on a 300K-token session is 3M Claude tokens spent on checking whether anything happened.
What I'm changing
Three changes, all dispatched to MiMo this morning (yes, through the same queue):- A
waitcommand that blocks until an order changes state and returns at once. It's read-only, so the orchestrator can run it over a mounted path from a VM. That replacessleep 595. - An automatic review as soon as an order's checks pass, run by Claude Sonnet in headless restricted mode. That means no shell, file reads confined to the review worktree and no MCP servers. The verdict is waiting by the time the orchestrator looks.
- A self-review section in every implementation prompt built from the five groups above, plus an order template with a mandatory "Failure modes" section. Cheap tokens make a self-check worth running.
Two of the three merged after two review rounds. The reviewer change took nine. It runs outside the sandbox, and for four rounds in a row a reviewer found another way a file planted in the repo could get executed through
PATH. In the end I dropped thePATHlookup instead of patching it again. That one order used 127M tokens, about $1.02 at MiMo list prices. At Opus prices the same tokens would have cost more than all 62 orders above did on MiMo.The aim is to get first-round merges from 39% to about 60% and take 20 to 25 minutes off a typical order. I'll post the before and after once there are enough orders to compare.
Limits
- One person, 14 repos, two weeks, mostly Python. Your mix of work will differ.
- The Claude timing comes from different tasks than MiMo's, so it says something about speed and nothing about quality.
- Rounds are counted from the worker's event logs. The wall-clock breakdown covers the 33 orders with complete timestamps.
- Prices are list prices checked on 25 October 2026: Xiaomi's rates as listed on OpenRouter, Anthropic's from its pricing page.
If you're splitting work between a planner model and a cheaper worker, measure the gaps between steps before you shop for a faster model. I do this kind of measurement as a routing audit ([email protected]). If you've measured your own split, I'd like to compare numbers.#ai #llm #agents #productivity #software #coding #development #engineering #inclusive #community
I routed two weeks of coding to a model 25x cheaper than Opus. The model wasn't the slow part. -
One code review comment can wreck your whole afternoon, even when the comment is about a variable name. I will show you why that happens and the small daily rep that takes the sting out of it.
Quick background so you know where this comes from. I'm Christopher. I spent nine years in the Marine Corps, then moved into software and data engineering, including edge computer vision work for the Air Force on a project called Athena. In the Marines, feedback came loud, blunt, and in front of everyone. You learned fast that a correction was about the task, not about your worth. And I still caught myself in tech reading "nit: this could be simpler" and spending twenty minutes wondering if my lead thought I was bad at my job.
If that is you, you are not thin skinned. Your filter is reading the wrong message.
What you actually hear
Your reviewer writes: "Can we pull this into a helper?"What you hear: "You don't know what you're doing, and now everyone on the PR can see it."
That gap is the whole problem. The comment is about twelve lines of code. The story you tell yourself is about you. So you either get defensive and argue every point, or you go quiet, accept every change without thinking, and open your next PR with your stomach in a knot.
Why it hits so hard
I don't believe you build confidence. I believe you unhide it. The same developer who calmly walks a junior through their first rebase can feel crushed by a two word comment on their own code. The skill is still there. What changed is how exposed the moment feels.Your brain runs a selective filter called the reticular activating system, the RAS. It decides what gets your attention, and it learns from what life fed it. Maybe a teacher marked your work in red in front of the class. Maybe an early reviewer was genuinely harsh. Maybe you just grew up where mistakes meant trouble. Your filter took notes: being corrected equals being exposed, and being exposed equals danger.
So now, when a review comes in, the filter skips past the five approvals and the "nice catch" and locks onto the one comment that sounds like criticism. You cannot logic your way out of that. Telling yourself "it's just code" bounces off, because the filter does not run on logic. It runs on experience. So you feed it new experiences.
The daily rep: compliment a stranger
Once a day, give one genuine, specific compliment to a stranger. Great shoes. Love that coffee mug. Your dog looks thrilled to be here. Say it, smile, keep walking.I know that sounds unrelated to a pull request. It is the same muscle at low stakes. You are putting yourself out there with zero control over how it lands, while your body braces for a bad reaction. The first few will spike your heart rate. Then they stop.
The rule that makes it work: the win is the action, never the result. They ignore you, win. They say thanks, win. You only managed to think it, still a win. If you only count the ones that go well, you are training the old program that says putting yourself out there only counts when people approve. The rep is the point.
That is rung one of a short ladder:
- Genuine compliments to strangers
- Compliment plus a quick "how's your day going?"
- A longer conversation, no agenda
Move up only when the current rung stops spiking you.
Bring it to your PRs
Once the stranger reps feel easy, run the same reps on your reviews.Read it twice, answer it once. First read is for your feelings. Let them happen. Second read is for the code. Only reply after the second read.
Rewrite the comment in plain words. "Can we pull this into a helper?" becomes "This logic repeats, a helper would make it easier to test." Strip out the tone you imagined and keep the request. Most of the sting lives in the tone you added.
Ask one question per review. "Do you mean a shared helper or one local to this file?" A question turns a judgment into a conversation, and it shows you are engaged instead of hiding.
Push back once, politely, when you disagree. "I kept it inline because it's only used here. Happy to change it if you feel strongly." That one sentence is a huge rep. Each time it goes fine, your filter learns that disagreeing is safe.
Say thanks in the thread. "Good call, that's cleaner." Closing the loop in public trains your brain that feedback ends well.
What Athena taught me about feedback
On Athena, edge sensors ran a local model. When it flagged something significant, it sent context from before, during, and after the event back to AWS so the models could be retrained. Nobody treated that data as an insult to the model. It was the only way the model got better.Code review is the same loop. A comment on your PR is not a verdict on you. It is training data for you and for the codebase. The developers who grow fastest are not the ones who get the fewest comments. They are the ones who can take a comment, use it, and ship the next PR without flinching.
You already know how to take feedback. Old training just taught you to hear it as an attack.
Your mission this week
- Compliment one stranger a day for seven days.
- On every review you get, read it twice before you reply.
- Ask at least one clarifying question on a review.
- Push back politely on one comment you genuinely disagree with.
- Thank a reviewer in the thread every time a comment makes your code better.
Then tell me in the comments: what is the review comment that stuck with you the longest, and was it really about you or about the code?#career #mentalhealth #codereview #beginners #software #coding #development #engineering #inclusive #community
You take code review comments personally. Let's fix that.