c.im is one of the many independent Mastodon servers you can use to participate in the fediverse.
C.IM is a general, mainly English-speaking Mastodon instance.

Server stats:

2.9K
active users

#elasticsearch

11 posts11 participants0 posts today

Продуктовый поиск на ElasticSearch

Из‑за санкций, эластик многие стали отвергать. Единственное, что его спасает, это RAG'и шмаги. Я первую работу выбрала в поисковом отделе, для меня это было гораздо «сексуальнее», чем datascience, с которым я была уже знакома. 8 лет назад начала изучать эластик. Наработалась «интуиция» на его функционал, так как с нуля я делала поиск для других компаний несколько раз. Поиск текстовый, GEO поиск, подсказки для поиска (suggester) и с МЛ'ем которым майнились синонимы или кластеризовались запросы. Открыть

habr.com/ru/articles/895606/

ХабрПродуктовый поиск на ElasticSearchВведение Видение Подготовка (ака тоже лайфхак) Лайфхак (без кода пока что) Метрики Введение Из-за санкций, Elasticsearch многие стали отвергать. Единственное, что его спасает, это RAG'и шмаги. Я...

Reading @joannastern article about giving up google for a chat bot, i have pretty much done exactly this.

I use google when I need to get an overview of something specific that I know nothing about (what exactly does #ElasticSearch do?).

When i already know what the answer looks like (how do you add a relationship in #Neofj using #ython?) I always use a chatbot.

When i get a giant stack trace, i drop that right into a chatbot.

wsj.com/tech/personal-tech/goo

Разбираем Pino: как работает самый быстрый логер для Node.js

Pino — один из самых быстрых логгеров для Node.js, но как он устроен внутри? В этой статье мы углубимся в низкоуровневые механизмы: разберём, как работают транспорты, чем отличается multistream от pipeline, и как настроить логирование без потери производительности. Вы узнаете, как использовать child loggers, маскировать данные с redact и избегать типичных проблем. Этот материал для тех, кто хочет понять Pino на фундаментальном уровне и применять его с максимальной эффективностью.

habr.com/ru/articles/894646/

ХабрРазбираем Pino: как работает самый быстрый логер для Node.jsЛоготип Pino Логи там, логи здесь... Логи — везде. Каждое современное приложение нуждается в логировании. В каждом языке есть свои фреймворки и библиотеки для сбора логов...

every now and then I'm like "I want to learn a new metrics/timeseries db" ... then I start doing that and I'm like "ughhh, I've been doing this in at least 6 different ways over the last 15 years, why do it again"

I really miss Graphite

I am responding to another email chain.... Thinking to myself... I wish more of my friends/family are on mastodon. This email is exactly how I use mastodon. Same with a text message.

Only things missing are:
- Encryption on private toots.
- Ability to send files like a pdf.
- Groups. A group of contacts that can be saved as one name. Each contact can receive/respond/remove own contact from group.

# Surprise Async ! (in python)

9pm and I am finished with my work.

This is fine, they think i'm old and slow. I think I make
way less errors especially the first time.

I'm not complaining (well obviously i am :P, but i don't
expect better i guess i mean).

Today's slow down was "surprise all that python you wrote needed
to be async" :)

We should be using golang here. I mean that practically given the
programmers we have. Actually idk ... I just assume golang
will work well on Azure stuff (mostly Azure Functions).

That's where the surprise async comes from. I'm still getting
a feel for regular sync vs. async. I'm more comfortable with it
now, meaning I have a decent idea when to use it. Usually for
me that means following whoever made that decision before me.
I think that's where the "async is a virus" idea comes from.
Once you are using it in a section, it all has to be async.
Which can not always be obvious in shared code.

I had to duplicate some elasticsearch_dsl code to work
in both sync and async. I really tried to use multiple inheritance
to not have to duplicate a Document class's field definitions.
But 1. I don't really ever want to use multiple inheritance and 2.
it wasn't working anyways.

If I knew this was going to happen I would have used regular
long form elasticsearch code, because it's driven by just dictionaries
and those could definitely have been shared commonly between
the sync and async bits. But I was too far down this path [0] so
I went with duplicating one of the table/index classes with
some comments saying they need to stay in sync.

If you are following this far ... well wow, I think that's awesome.
Thanks. The sync class is just to initialize an elasticsearch index
once. Like it will run once in each environment and then proceed
to be a no-op for the rest of eternity. Otherwise it's all async and
await funny business.


[0] Too far down the path of using the higher level elasticsearch_dsl
python library. Which wraps the otherwise normal http(s) requests
in python code with DSL like search stuff. The regular elasticsearch
python (non _dsl) is just a thinner layer over REST style GET/POST
JSON calls. So it's more verbose, but also kinda nice because you
can see you are just making REST calls.


#techpost #python #async #elasticsearch #azure

I know I give #Elasticsearch a lot of crap, but I've got a sales meeting with the Elastic folks to this morning to talk about their SIEM offering (which I love) and I'm totally geeking out about it. I can't wait for my manager to see it and understand why I love it so much.

Unfortunately our last sales meeting was with Crowdstrike and they're pretty flashy and impressive so we'll see what happens.