Ivan Vergiliev

[Ivan Vergiliev @IvanVergiliev]

I wanted to brush up my front end skills so I implemented a clone of 2048 with @reactjs: http://t.co/eJhcJXfm6c . It's touch-enabled!

[Ivan Vergiliev @IvanVergiliev]

@GergelyOrosz Assuming you still review the generated code, I’ve found that “the agent produced garbage ===> the surrounding code is likely hacky and not well-structured” is generally a pretty good heuristic. If the agent can’t figure it out - refactor.

[Ivan Vergiliev @IvanVergiliev]

@clare_liguori @JillianKozyra With some light `gh` scripting, you can easily get reasonable PR support too. I have `jj_pr <foo>` and `jj_prs c1...cN` to push a whole stack.

I personally use the "strange" commit per PR model once I switched to jj 😅 But I've done multiple commits occasionally - also viable

[Ivan Vergiliev @IvanVergiliev]

@clare_liguori @JillianKozyra - mutable commits → you can directly edit a PR earlier in the stack
- auto-rebase → whole stack gets updated automatically if you change the base or anything in the middle
- you can easily push the whole stack with a single command

[Ivan Vergiliev @IvanVergiliev]

@edpizzi I've definitely wondered in the past why Protobuf doesn't have a serialization format defined. Thanks for the inside perspective!

[Ivan Vergiliev @IvanVergiliev]

Re-reading the Google File System paper, one thing I notice is all the focus on concurrent appends and the use of files as producer-consumer queues. Most cloud storage services (AWS S3, even GCS) nowadays don't support appending at all; queues are offered as a separate product.

[Ivan Vergiliev @IvanVergiliev]

@edpizzi I wasn't familiar with recordio - I guess I've only ever worked with individual protobuf records. Interesting how file system design can impact the internals of a file format.

[Ivan Vergiliev @IvanVergiliev]

@DeveloperKhare The only other things was the fact that our job was getting canceled shortly before the TaskManager died. This was part of the reason I spent more time considering possible issues caused by the job cancellation itself.

[DeepCode @DeepCodeAI]

https://t.co/nlnv8x3D0Y

[Ivan Vergiliev @IvanVergiliev]

@DaskalovBoris Do you know if Google still uses GFS in that manner, or if they have something internal in the spirit of Kafka now?

[Ivan Vergiliev @IvanVergiliev]

@DaskalovBoris You can certainly do that with a tool like Kafka - just set the retention period to an appropriate value. Seems like AWS Kinesis also supports that. "Distributed log" is probably a more accurate term for this category of systems though.

[Functional Works @Functionalworks]

A great article for you: 📜

How We Found a Missing Scala Class - by @IvanVergiliev

"Here’s a fun tale of debugging that starts off with a common alert nagging our on-call rotation..."

https://t.co/ljspUTJalf

[Functional Works @Functionalworks]

Debugging may stem from a common alert nagging your on-call rotation 🐞but could turn into you knowing a lot more about dynamic tracing in the #JVM 💻

@IvanVergiliev explains how @heap ended up finding a missing #Scala Class

https://t.co/dlVzzRNV1X

[Signify Technology @Signify_Tech]

We bring you our recommended Sunday Reads, this week we hear from @adamwarski, Dean Povey, @IvanVergiliev, Lettier, @SethTisue & @AshNuq covering #Scala, #Haskell and more. Happy reading! 📖 📚 https://t.co/NJuLtEPjl3

[Scala Times @ScalaTimes]

Scala Times Issue #240 - https://t.co/aUteI5sZRC by @47deg @apnylle @cbirchall @DevInsideYou @IvanVergiliev @mlangc @Odersky @ScalaUpNorth @shanielh @Signify_Tech & others, thx!

[Janet Todorova @janettodorova]

@IvanVergiliev talks about #scaling at @HackSoft_ // not just reading about scaling on @newsycombinator but actually doing it // cc @leanplum @heap 🤖🚀👏 #hackconf2018 #tech #sofiastartup

[Ivan Vergiliev @IvanVergiliev]

@cgenco @patio11 Ignoring the `setState` call, isn’t this infinite recursion? Seems like you’ll keep calling `updateItem` forever.

[Thomas Boussekey @ThomasBoussekey]

Pgbouncer internals explained! https://t.co/cSaKL3NKsn

[Ivan Vergiliev @IvanVergiliev]

Ever wondered how a connection pooler works under the hood? I did some research into PgBouncer internals a few weeks ago and summarized it here: https://t.co/y9qQIcQkYF #PostgreSQL

[Ivan Vergiliev @IvanVergiliev]

@izomorphius Първоначалната ми реакция, след като прочетох първото ти изречение и Tweet-а, беше, че това е втори от 182 tweet-а...

[Ivan Vergiliev @IvanVergiliev]

CPU instruction pipelining is a bit like Kanban - you move tasks between different stages, have upper limits per stage, and are blocked by the slowest stages. #UselessAnalogies