JShell project for Tj-Bot!

[sterile crest]

It can become spaghetti if we don't approach correctly

[compact holly]

but grpahana for us would be a.. bonus feature

we survived without it

it that subsystem fails for whatever reasons, we will be fine

if it's designed in a modular way like that

[sterile crest]

You survived without it, I can't

I hate that I couldn't find where jshell was invoked manually in the server

I fr checked bots, spam, chitchat, questions

Maybe I'm blind, I couldn't find it

[compact holly]

yeah, you don't have diagnostic tooling

you don't have that one faulty session to investigate

you are not even aware where it happened

and why

what was the input.. etc

[sterile crest]

We don't even log it in the logging channels that somebody invoked a command:(

[compact holly]

this can all be improved

for example sessions that failed, errored out, could be perserved for 14 days

[timber mirage]

wdym, how could you not find it ?

[sterile crest]

Yes kinda

[compact holly]

so if nothing else, you can manually play with them

[sterile crest]

Skill issue prob

[timber mirage]

Let me find what you want

[compact holly]

you kinda want everything, because currently it's hard to triage and reproduce bugs

and investigate

because it's just random sessions being spawned and killed

so if there is issue, you only have logs

logs are improved now tho, right?

[sterile crest]

#infra-status message how about the most recent

[compact holly]

that's easy, pick one that is not active

[sterile crest]

So when are we supporting different runtimes

I wanna run Kotlin

[timber mirage]

wdym

[sterile crest]

Or groovy

Or ruby

I wish 😦

Tbh

I'm gonna suggest something crazy

But it's rewriting jshell

So we can support everything

[compact holly]

i mean, project like that would...

[sterile crest]

Now that we have it working on the vps

[compact holly]

we could make it commercial :p

[sterile crest]

Less than a week, I have a solution

[compact holly]

easy money

but it's hard work

[sterile crest]

It's not

[compact holly]

it always looks easy, until you realize how many complexities there are

but it would be a top product on the market

it runs any code

[timber mirage]

There are two entry points for calling jshell in tj bot :
https://github.com/Together-Java/TJ-Bot/blob/develop/application/src/main/java/org/togetherjava/tjbot/features/code/EvalCodeCommand.java
https://github.com/Together-Java/TJ-Bot/blob/develop/application/src/main/java/org/togetherjava/tjbot/features/jshell/JShellCommand.java
If you want to collect stats, you could add that in class
https://github.com/Together-Java/TJ-Bot/blob/develop/application/src/main/java/org/togetherjava/tjbot/features/jshell/JShellEval.java
Since all calls pass by here.
@sterile crest

[sterile crest]

I have the solution in my head

[compact holly]

and it's integrated in discord chat, on a website

on any platform

[timber mirage]

So

[sterile crest]

I wish I had my own discord bot in this server 😦

[compact holly]

you can send output to email or your messaging server of choice

[timber mirage]

If you want to add graphana

just add code in JShellEval in TJ-Bot

and you are done

easy @sterile crest

[compact holly]

or in your terminal, like universal repl for every language

[sterile crest]

Okay you're misunderstanding how it works

But yes it's easy

[sterile crest]

No man

/jshell lang=ruby code=whatever

[compact holly]

your head must be so big right now

[sterile crest]

Then for the docker part, just use the official docker image for the language

And just <command> <code>

And spit out the results

[compact holly]

yeah, whole AWS paltform is basically just few 'justs'

just spawn a server, just spawn an image and run it for 30 seconds

[sterile crest]

But I didn't mention AWS

We can use our infrastructure

[compact holly]

i mean, every software is 'just'

[sterile crest]

My brain is just bigger than yours

[compact holly]

most shit we work on are basic CRUD apps, like websites

you get some text, you pull something from the database, you send it back

easy

[sterile crest]

Yeah

Easyyy

Peasy

Lemon

Squeeezyyy

I'm weeezy

Like breezyy

Now I'm rapping

[timber mirage]

Problem is that jshell is a REPL, I coded it in a way to expect a REPL
So we can only use languages that have a REPL

[sterile crest]

[timber mirage]

unless we do some magic

[sterile crest]

Yeah magic called abstraction

[timber mirage]

but yea it could work

you would only need to add a new wrapper

for each language

[sterile crest]

Exactly

You got itt

You understand

I want my own bot added to TJ 😦 we can call it TJ community commands Bot

[timber mirage]

and the thing is that the wrapper doesn't use json or any complex format, so you don't even need any lib in the wrapper

[sterile crest]

Where people submit their own fun commands

I'm gonna suggest it

Where's the channel

#server-suggestions

[compact holly]

yeah, in theory it's all simple and cool

because we built our own in-house aws lambda

and we can run anything on it

[sterile crest]

You're acting like I can't POC the idea in under day

[compact holly]

that's why we have api, thinking about the future man (that you had doubts about, and thought was a dogshit idea)

[sterile crest]

#server-suggestions message

[sterile crest]

It is dog shit

Unless we're taking outside of TJ and making money

@timber mirage btw the way graphana works is...

We pull a docker image and that will start a web server

Then from any of our services we just push the metrics to a data source (e.g. promethius or a sql database)

And on the graphana website that it spun up, we can see the raw metrics that we collected and start making graphs for it

I.e. a new web page that contains stuff

E.g. "jshell" then all the stuff

And "TJ slash commands", "TJ errors" etc etc

Or however you want

We have flexibility on how we visualise the data

If I remember correctly, the graphs are just json files e.g.

{
tile: {
name,
x
y
data
aggregation
}
}

Some shit like that

[bright lark]

Springboot admin can do the monitoring stuff

With so much less efforts

[sterile crest]

uh no

[bright lark]

Whatsp

[timber mirage]

@bright lark any advancement ?

[bright lark]

hi

not yet got some pressure recently at work

i'll try to check asaic

[timber mirage]

alright

[bright lark]

yesterday when i replied i was in bed

unable to sleep and got multiple notifs from u

so i wanted to join by phone

(i wasn't in front of laptop)

[timber mirage]

Take your time

[bright lark]

right, so im gonna do the fixes and what then ?

if we agree on a final approach, i'd probably go to the next usecase to write test for it

(btw im gonna explain what i did in readme)

[timber mirage]

yes please

[bright lark]

alr thanks for ur trust

[bright lark]

hi again and sorry i couldn't find enough time for pr review today, maybe tomorro stay tuned

[bright lark]

hello @timber mirage ! please answer me here https://github.com/Together-Java/JShellPlaygroundBackend/pull/50#issuecomment-2347256728

i mean on discord

to move faster

thanks in advance

[timber mirage]

26 lines to test a simple 2+2 and you call your test simple ? it's also extremely redondant
for the second point, I am asking to test for state

[bright lark]

I removed the constants

So u want to put the webtestclient chaining inside utility method?

So it becomes testEval(args)

Right?

[timber mirage]

I gave an example

[bright lark]

The test for state is also there

But the example u gave is about a list

[timber mirage]

where ?

[timber mirage]

a list ?

[bright lark]

boolean isValidResult = result.snippetsResults()
            .stream()
            .filter(res -> res.result() != null)
            .anyMatch(res -> res.result().equals(TEST_CODE_EXPECTED_OUTPUT));

This checks if result has 4

[timber mirage]

why don't you just check there everything is correct ?

why can't you do what i said ?

i don't understand

[bright lark]

I did

[timber mirage]

?

[bright lark]

Ive checked tht everything is correct in the isValidResult variable

[timber mirage]

but that's not what i asked

I asked to simplify the tests

and add a test to test state

[bright lark]

ah Thats another test

Ill do it when we merge the pr

[timber mirage]

no

do it before

[bright lark]

Why no?

[timber mirage]

because we have no idea if your approach works with state

[bright lark]

U mean reuse session

?

[timber mirage]

yes

[bright lark]

My approach in the current test works just fine

No need to go faraway from this basic scenario

When i add the second test, i may do that

We agreed on setting one test in my pr, remember?

[timber mirage]

we need this second test

[bright lark]

Hmm

I didn't think about it

But i still dnt see the reason to block pr for tht, the wrapper container is shutdown when the test is finished

[timber mirage]

?

what do you not understand ?

[bright lark]

I mean we can merge the pr

Without adding another test

Anyway if u insist then i dnt have choice

Gotta go sleep now sorry, ill get back tomorro

[timber mirage]

we can't because we don't know if it will work if you don't do this test

[bright lark]

Anyway if u insist then i dnt have choice

[timber mirage]

I mean, try to understand why it is needed

[bright lark]

i know it's needed !

but i just want to set exactly 1 test

in this pr and next we do others

as agreed previously

[timber mirage]

You misunderstand, it is needed for this pr, it's not a random test

[bright lark]

this pr is about testing one scenario in 1 endpoint

the test u want me to add is a second scenario

and ok np

[sterile crest]

Firas is right though

Unless the test you're requesting is directly related to a change he has made

[timber mirage]

It is

[timber mirage]

No, the pr is about creating the infra to make tests

[bright lark]

alright

bring it on

https://tenor.com/view/pokemon-ash-gif-17972604

i'll take a look tonight

[timber mirage]

And the reason we need tests is to verify that the infra works

[bright lark]

huh

i didn't say no to that

i assume all integration tests are obviously relevant to infra

but nm

[bright lark]

Hi again! @timber mirage the next test is about using same session id (e.g: "test") used for "2+2" but with java expression right ?

[timber mirage]

store the 2+2 in a variable, and reuse this variable in the second call

[bright lark]

testEval(testEvalId, "2+2", "4");
testEval(testEvalId, "2-1", "1");

ah !

that means i must get same result again ?

[timber mirage]

testEval(testEvalId, "int a = 2+2", "4");
testEval(testEvalId, "a * 2", "8");

@bright lark

if you are correctly using the same session, the variable "a" should exist

[bright lark]

let me check that out

it works ! ✅

https://tenor.com/view/rasputin-fifidonia-dance-bailar-alegria-gif-20386526

[timber mirage]

also can you do it the way I gave you in the first point ?

[bright lark]

i did like this

[timber mirage]

but what if you want to test anything else than the result ?

[bright lark]

like wat ?

[timber mirage]

Anything that is in JShellResult

[bright lark]

wdym

the result is as expected

the response is success

[timber mirage]

What do you not understand ?

[bright lark]

u want me to check something else

[timber mirage]

but you didn't check any of the other fields, how can you know that ?

[bright lark]

the http response header is enough to tell if it successful or not

[timber mirage]

no it's not

please check everything

it doesn't even cost you more lines of code that what you currently have

[bright lark]

status ?

[timber mirage]

for example

[bright lark]

alr

ill add this and done

[timber mirage]

no

add everything

this should only take one minute

[bright lark]

this is weird

it's not about how much it takes

it's about the reason

[timber mirage]

because you can't just check one of the values and assume that everything is correct ?

[bright lark]

i mean result and status are obviously enough

but ok

[timber mirage]

it cost nothing to check everything

[timber mirage]

and no, it's not enough

[bright lark]

testEval(testEvalId, "int a = 2+2", "4", SnippetStatus.VALID, SnippetType.ADDITION);

testEval(testEvalId, "a * 2", "8", SnippetStatus.VALID, SnippetType.ADDITION);

[timber mirage]

no

first of you didn't test everything

and I gave you the solution

[bright lark]

all is left is the id

please just tell me what else

[timber mirage]

just do what I told you to do

[bright lark]

I think I DID

if u think that i didn't , then just tell me so i go to it straight

[timber mirage]

Your test is too complicated to test something too simple, you need to simplify it, create methods or classes for this so your test can be as simple as for example assertEquals(sendEval("2+2"), new JShellResult(...))

[bright lark]

ok

[bright lark]

assertThat(testEval(testEvalId, "int a = 2+2;")).isEqualTo(new JShellSnippetResult(SnippetStatus.VALID,  SnippetType.ADDITION,1,"int a = 2+2;", "4"));

assertThat(testEval(testEvalId, "a * 2")).isEqualTo(new JShellSnippetResult(SnippetStatus.VALID, SnippetType.ADDITION,2, "a * 2", "8"));

[timber mirage]

yes

wait

[bright lark]

alright we're don..

[timber mirage]

why not JShellResult

[bright lark]

🤷

the snippet comes from result

[timber mirage]

use JShellResult

please

[bright lark]

i think we'll not end

[timber mirage]

Why don't you just do what I asked days ago ?

Why do you try to hard to find a way to not do that ?

[bright lark]

dnt be mad

alright it's done

im gonna commit

now i'm writing some docs in readme

[bright lark]

@timber mirage all is done !

[sterile crest]

@timber mirage wanna setup graphana and promethius with me?

promethius is what we'll use as a data store and graphana is how we'll create pretty charts

but i'm worried we'll have to fk around with networking so maybe we can get something working locally first and then... think about how to deploy it with marko, without having the same issues we did with jshell

since we need to expose 2 things:

1. promethius to graphana and tj

2. graphana to the web

[timber mirage]

why not

[sterile crest]

I need to remember how to do this

I think TJ needs to export a website with metrics?

[bright lark]

why u dnt use spring boot admin

[sterile crest]

can you explain spring boot admin?

[bright lark]

it's a tool to follow deployed spring boot app

just add the dependency

and done

[sterile crest]

where in TJ are we using spring?

[bright lark]

i mean for jshellapi

as it's spring boot

nevermind

[sterile crest]

we can perhaps use the existing spring stuff in jshell to expose the metrics

  prometheus:
    image: prom/prometheus:latest
    container_name: prometheus
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
    ports:
      - 9090:9090

  grafana:
    image: grafana/grafana:latest
    container_name: grafana
    ports:
      - 3000:3000
    environment:
      - GF_SECURITY_ADMIN_PASSWORD=admin

(keeping this here so we don't lose it) - it's the docker-compose config

prometheus.yml ```
global:
scrape_interval: 15s

scrape_configs:

• job_name: 'prometheus'
  static_configs:

  • targets: ['localhost:9090']

• job_name: 'java-app' # You will update this to your app's details
  static_configs:

  • targets: ['<your-java-app-ip>:8080']

[timber mirage]

@sterile crest why do you try so much to make the aws solution comes into tj bot ?
what do you expect to do with it ?

[sterile crest]

where is AWS

you mean the PR?

Marko asked to see it, I only spent maybe 2 hours creating the PR and left it, idc if its merged or not

[timber mirage]

yes

[sterile crest]

but you know, 2 hours vs 6 months...

it's a big difference

[timber mirage]

you probably didn't take 2h to do it

[sterile crest]

it was 2 hours

less than 1 day, when I originally did the POC, it took 30 mins

remember, there's no docker, or any complicated logic

it's a simple eval API, that's it, no sessions etc

[timber mirage]

such api, I had one in late 2023

yes it took more time, but 6 months, it was 6 months waiting mostly

anyway

[sterile crest]

everything is a wrapper over this:

that's jshell

and then in AWS I am just OBJECT_MAPPER.writeValueAsString(JShellOutput); and AWS side is done

the rest was just the slash command

the PR looks big, but's mainly c&p from other projects I've done in the past

but again, idc about that solution now

though, with AWS we could skip promethius 😦

but we have to set that up now and all the complexity around getting metrics exposed

[cedar fog]

OTel's not that hard to implement in and of itself, but properly defining what to track is another matter.

And having enough people with the required rights.

[timber mirage]

anyway yes, tell me if you need help
but, why not setting up that on tj bot directrly ?

[sterile crest]

don't think you can?

[timber mirage]

wdym

[rotund kernel]

Good afternoon

I would like to join

[timber mirage]

Hello, wdym ?

[rotund kernel]

the project I mean

or it is already closed?

[timber mirage]

nono

[timber mirage]

try to install the project then

[rotund kernel]

okay just a moment

[rotund kernel]

installed

[vivid cliff]

no its not

[timber mirage]

then read the readme and try to run it
tell me if you have any question

[vivid cliff]

https://github.com/Together-Java/JShellPlaygroundBackend

this is the repo

u can check it out

[rotund kernel]

okay will reach you out

[rotund kernel]

thanks man)

[rotund kernel]

it's stuck on second command

how can i help you out?

[timber mirage]

wdym

[rotund kernel]

first of all, i got a problem with running JShellWrapper:run

and i tried to ./gradlew bootRun

and it doesn't work for me

stuck on 88% messaging
2024-09-14T18:45:51.920+05:00 INFO 12684 --- [pool-2-thread-1] o.t.j.service.JShellSessionService : Scheduler heartbeat, sessions ready to die: []

[timber mirage]

wdym, does it print this log endlessly ?

[rotund kernel]

yep

while at the bottom

Executing: 88%

<===========--> 88% EXECUTING [7m 20s]

and also about JShellWrapper:run

it is giving an exception requiring me to accept Terms&Conditions at gradle website

[bright lark]

👀

Never saw this

[bright lark]

Ah the problem is with env variables

follow The readme file

U need to 2 env variables with specific values to make it run

[bright lark]

U need to use comma

Rather than semicol

Between the variables

Thats how it works in intellij

[rotund kernel]

okay just a moment

[bright lark]

And dnt use spaces

[rotund kernel]

what about this

[bright lark]

Oh damn

I got confused

use semicolon

But without spaces

[rotund kernel]

okay

[timber mirage]

if you executed JShellAPI, that's normal i guess

[rotund kernel]

ahhh then okay

how much it actually takes to make a run

[bright lark]

Seconds ig

[rotund kernel]

JShellWrapper:run i mean

[rotund kernel]

it is going minutes for me

[bright lark]

U need to understand somthing here

U dnt have to run jshellwrapper by yourself

It is implicitly done by jshellapi

All u have to do is build the jshellwrapper img

And use post:eval endpoint

If u do, jshellapi communicates to docker and runs the img

Making it living for a period of time (session)

[rotund kernel]

okay thanks

[timber mirage]

@rotund kernel so ?

[rotund kernel]

ran it

but thinking about port 8080

Web server failed to start. Port 8080 was already in use.

Action:

Identify and stop the process that's listening on port 8080 or configure this application to listen on another port.

i couldn't be able to kill it

it is creating one auto

[timber mirage]

are you already using port 8080?

[rotund kernel]

yep

will change the config file

after will set it to default

[timber mirage]

right

mention me if you managed to make it work

or if you have any question

[rotund kernel]

okay

[timber mirage]

@bright lark I reviewed your pr
Basically, you are transforming the readme into a report showing what you did... that's not what a readme is for
And some other changes
Ah and specifically one of them, is it possible to make so tests use a custom config file ?

[timber mirage]

any advancement ?

[sterile crest]

I agree with Ala's comments on the README

[bright lark]

hello

im checking ur comments

it looks like u didn't like

i tried to do some stuff as i thought about newcomers (new learners)

Ah and specifically one of them, is it possible to make so tests use a custom config file ?
sure

you'd like to set a config record and properties like the one under src/main ?

[timber mirage]

Basically setup a custom application.yaml but for tests
And the idea is that different tests can use a different one

[timber mirage]

I explained this in the readme, this is at best the wrong place.
And there are many things you put that won't help anyone anyway
What's the point of copy pasting the build.gradle, can't people just check it out ?
What's the point of copy pasting gradle rules, like file structure ?
Why do you need to explain each line of code you did in the readme ?
What's the point of saying that you, and us did this work in a random place in the readme ?
You are supposed to make doc for this, javadoc or equivalent for this
You are just making a report for a school project.

Also remember that readme purpose is to explain people how to use the project, not how to explain how it works.
Seriously, if I really did what you did but for the whole project instead of your test, the readme would be 10k lines long, it would be unreadable.

[bright lark]

I see. Well it's cuz u said u dnt know about gradle and my changes

So i thought i have to write some friendly hints to help unfamiliar contributors to understand

Somebody may not know about injecting the image name from root build.gradle file to subprojects

Also i tried to let know that the test class is about integration test and that it depends on additional configs in build.gradle file

Now i dnt have much time to discuss sorry. Will come back soon

But thanks for ur feedback

[sterile crest]

You can still have the documentation that you want just not on the parent README or in file comments that don't make sense

Instead, what we can do is create a docs folder that contains all our documentation as .md files

[timber mirage]

or create a wiki, like tj bot did

well, that's basically the same solution

[sterile crest]

Yeah even a docosaurous module

[timber mirage]

but even then, many of the things you wrote shouldn't be there

like there is no point to write gradle or maven rules

like when you said where the tests should be located

[sterile crest]

Maybe it's better suited for some contributing guidelines but whatever the case, there should located somewhere more specific for docs

[timber mirage]

and you need to make a difference between choices that you did and that you want to remember by writing them out
and actual explanation about how the project works

[timber mirage]

first is only useful if you need to edit that specific line of code
second is useful for everyone

[compact holly]

there should be just an abstract overview how testing works, with simple example

[timber mirage]

the example is quite literaly the test itself

[compact holly]

basically a broad guide on how to write tests on our project

[timber mirage]

it's still out of the scope of the main readme

[compact holly]

obviously ^^

[sterile crest]

He even said "thanks ala & wazei - Firas" in the README

It was cute

[bright lark]

Hello

Use Objects.requireNonNull instead of abusing Optional.

i wonder why u say that @timber mirage

[timber mirage]

Ignore this one since you won't need it if you fix the previous comment

[bright lark]

i assume your review was too strict

but i like it

[fossil badge]

invalid link, is this a bot to teach java?

[timber mirage]

Ah, it was the test server, now that it was deployed, it's useless to give it anyway.
And no, it's just jshell

[stray wren]

Alathreon's result

Snippets

Snippet 23, VALID

2+2```
jshell> `4`
## System out
[Nothing]

[timber mirage]

this

[fossil badge]

🤯

[bright lark]

[stray wren]

bigigga212's result

Snippets

Snippet 23, VALID

Random random = new Random();```
jshell> `java.util.Random@67117f44`
## System out
[Nothing]

bigigga212's result

Snippets

Snippet 24, VALID

random.nextInt(20) + 1```
jshell> `19`
## System out
[Nothing]

[fossil badge]

🤯

[timber mirage]

you know you can use nextInt(1, 21)

[fossil badge]

i donut

[timber mirage]

try it 🙂

[stray wren]

bigigga212's result

Snippets

Snippet 25, REJECTED

nextInt(1, 21)```
## [WARNING] The code couldn't end properly...
Problematic source code:
```java
nextInt(1, 21)```
Cause:
The code doesn't compile:
cannot find symbol
  symbol:   method nextInt(int,int)
  location: class 

## System out
[Nothing]

[timber mirage]

random.nextInt(1, 21) I mean @fossil badge

[stray wren]

bigigga212's result

Snippets

Snippet 26, VALID

random.nextInt(1, 21)```
jshell> `2`
## System out
[Nothing]

[fossil badge]

very cool

is it safe from something silly like .getLocalHost()?

[timber mirage]

you can try to break it if you want #bots

[timber mirage]

@fossil badge did you test it ? 🙂

@bright lark any advancement ?

[bright lark]

Actually not im in a terrible pressure at work these days

But gonna check soon

[timber mirage]

alright

[bright lark]

Tbh your review on my pr was legendary

It helped me a lot at work

So i reviewed some prs same way and it gives precious results

[bright lark]

@timber mirage i think about removing some content in the readme as u suggested

and put it in the PR description instead (not everything and briefly)

is that ok ?

[timber mirage]

I mean, you probably want to delete everything you added

since none of it is useful for an user

[bright lark]

huh

i thought automated tests aren't for user

[timber mirage]

yes ?

[bright lark]

the documentation is to help other developers to understand the approach

[timber mirage]

readme isn't documentation

it's user manual

[bright lark]

hmm

do i add a contributing.md ?

[timber mirage]

it's not documentation either

you would need to add a wiki

which

isn't priority for now

[bright lark]

alright but i'd put some details in the pr description about the approach

ok ?

[timber mirage]

in the pr itself ? you can
but remember, don't put things we already know or we can easily find, it's pointless to say where each file is, it's pointless to show code
and it's still not documentation
you don't explain to the programer how they should do it
you explain which challenegs you faced and the choices you made
this would have a place in the pr so we know how this was done

@bright lark actually, probably better to put such info in the issue, not the pr

so the issue has all info regarding the choices you made to make it work

[bright lark]

alright sounds good

https://github.com/Together-Java/JShellPlaygroundBackend/pull/50/files/3de55c67502851513fc53fb47f1cddf645090484#r1767481909

can u answer me quickly on this ?

[timber mirage]

No, I'm cooking

In one hour or two

[bright lark]

lmao

ok

ive answered on GH anyway

[bright lark]

no need i got the answer in the review !

https://github.com/Together-Java/JShellPlaygroundBackend/pull/50/files/3de55c67502851513fc53fb47f1cddf645090484#r1760314383

[bright lark]

hello folks

fyi, im about to resume my pr review fixes

[bright lark]

@timber mirage ive pushed changes

would you like to take a look pls

[timber mirage]

@bright lark any advancement?

[bright lark]

hello

what ?

ive finished

im waiting from someone to wrap up

[timber mirage]

Ah did you check the reviews I made?

[bright lark]

i think ye

[timber mirage]

Ah, I'll check that

[bright lark]

ive just made a new update

[timber mirage]

@bright lark I made a review

[bright lark]

there are some points you'right about

however ive answer u about others

[timber mirage]

Alright, replyied

[bright lark]

i see to put 2 configs : common and specific ones

for dev and prod: merge common + devconfig

for test: merge commong + testconfig

[bright lark]

https://github.com/Together-Java/JShellPlaygroundBackend/pull/50#discussion_r1775539621

u said that we'll fix that later

https://github.com/Together-Java/JShellPlaygroundBackend/pull/50#discussion_r1773629339

[timber mirage]

fix jshell name yes, but not the variable you just added

[bright lark]

why we dnt fix both ?

it doesn't make sense to keep one unfixed

[timber mirage]

You have to fix the one you added

And for the other one

it's not in the scope of this pr

and will cause many problems

because all the config files will need to be updated

[bright lark]

as u wish

https://tenor.com/view/as-you-wish-bowen-yang-saturday-night-live-your-wish-is-my-command-yes-sir-gif-23427642

[timber mirage]

In case you didn't understood

You added something

it's wrong

I am merely asking you to fix it

[bright lark]

wdym

[timber mirage]

you added a variable

the variable doesn't follow conventions

i am simply asking you to fix it

[bright lark]

but i think that properties aren't meant to follow camelCase!

when i see predefined properties in docs, all are lowercase

[timber mirage]

they are made to match the config class, if they match the variables in it, there is no doubt about how it will work

[bright lark]

they can follow the variables in config class even in lowercase

like i did

[timber mirage]

still

and even if

you have to follow the conventions of a project

that's the most important thing

if a project started with a particular convention

then you must follow it

[bright lark]

right

as for readme

i m going to just write the command to run tests

is that ok ?

or remove the section

[timber mirage]

Just remove the section I guess, we could add this later

[bright lark]

alright

so is this ok jshellWrapper-imageName

or i remove that - ?

nevermind

[timber mirage]

yes remove it

[bright lark]

another point please, ${jshellWrapperImageName} i m gonna change it

to uppercase with underscore

cuz it's like env variable

so i becomes ${JSHELL_WRAPPER_IMAGE_NAME}

jshellWrapperImageName: ${JSHELL_WRAPPER_IMAGE_NAME}

LGTM

[timber mirage]

yes do that please

[bright lark]

what do u think about this ?

@timber mirage

[timber mirage]

only use hardcoded config, it's better i think

[bright lark]

ye but i think about extracting some configs from Config

and put them in use in dev and runtime

and create a separate config for test that extends Config

in other words i'd have:

class Config extends CommonConfig {}
class TestConfig extends CommonConfig {}

[timber mirage]

no

[bright lark]

to avoid redundancy

[timber mirage]

there is no redonduncy at the fist place

you are the one adding it

just create a custom config object for the test

[bright lark]

why would i define the evaltimout twice?

as for the session it could customized

[timber mirage]

because it may be different ?

[bright lark]

but that will not make test consistent

[timber mirage]

?

[bright lark]

i think that integration test must follow the real scenario

if i set timeout in test lower or greater than the one used at runtime

[timber mirage]

they can't

we have a 30min timer

we can't make so the test takes hours

[bright lark]

ye i agree to make that shorter

so i think that some configs need to be customized between test and runtime

[timber mirage]

some yes

[bright lark]

but not some others

[timber mirage]

but we don't know which one

we just know that at least some need to be modified

[bright lark]

alright, so it would be a good idea imo to keep common ones centralized in one place

[timber mirage]

no

[bright lark]

and create extensions

[timber mirage]

because we can't know which one will be common

[bright lark]

we can decide

for example the session timeout

(it could be only this one)

😃

[timber mirage]

i could

or it couldn't

[bright lark]

we can start with it

and review others

[timber mirage]

also your approach is just worse

you are saying that you reduce duplication

but you are not doing that

you are doing the opposite

especially when the original solution is one line long

[bright lark]

i dnt see what u mean can u elaborate?

[timber mirage]

on what ?

[bright lark]

you are doing the opposite

i dnt see how's tht

[timber mirage]

the current solution is one line long

you are asking, to replace the record with three classes

[bright lark]

no

[timber mirage]

so around 150 lines of code

[bright lark]

you asked to create a copy from Config

[timber mirage]

new Config(foo, bar)
that's one line

[timber mirage]

no

well

kinda

it depends

either there is a test config

or we just new Config

[bright lark]

this looks like a violation to properties

well

as i said earlier "as u wish"

[timber mirage]

both have advantages

the problem of having a second config

is that we can't customize it

and so our tests will be more limited

so it's not a choice

it's weither or not we know what the tests will contain

[bright lark]

we can

[timber mirage]

how ?

[bright lark]

it's valid to add a properties file under src/test/resources

[timber mirage]

no

that's not what I mean

what I mean is that we can't select different config for different tests

if we use a config file

[bright lark]

we can

with profiles

@ActiveProfiles

[timber mirage]

but then it just makes things even more complex

well

I guess we could got for a single config file in resources

then we will change it later if needed

[bright lark]

that means no inheritance

configs are isolated

[timber mirage]

yes that's the point

and you shouldn't use inheritance for this

it's just a bad idea

[bright lark]

Ok

[timber mirage]

@bright lark hello, any advancement ?

@sterile crest @compact holly hello, I just made a pr to update jshell to java 23 : https://github.com/Together-Java/JShellPlaygroundBackend/pull/54
I have a question, is it possible to make so it is automatic for the next versions ?

[sterile crest]

You can write pure java code inside your build.gradle

So write a method e.g. int getVersion() and use it

[compact holly]

you don't want anything serious automated

[sterile crest]

Nope, but if we had Canary deployments....

It would be perfect

[timber mirage]

I just want that it automatically create a pr

[sterile crest]

But that's too much effort to teach because it's also doing the work

[sterile crest]

You can create a GHA

[timber mirage]

Yes that's what I am asking, but how ?

I have no idea how GHA works

[sterile crest]

How did you learn java?

You read books? YouTube? Etc right?

Go figure it out

[timber mirage]

Can't you teach me ?

[sterile crest]

Hmmmmmmmmmmmmmm

@compact holly should I teach him or let him figure it out?

Actually

Teaching is a big job

[timber mirage]

Or at least, can you tell me where I can find the code that create those auto pr for dependencies ?

[sterile crest]

Look inside the .github folder

[compact holly]

dunno, it's your time

[sterile crest]

Dunno has the word no, sorry ala

[compact holly]

but actions have docs, there are plethora of already made actions you can just use with a single line

[stray wren]

TJ-Bot

Does "Dunno" have the word "No" ?

Yes, "Dunno" does have the word "No" in it. "Dunno" is a colloquial contraction of "don't know," and it contains the word "no" within it. The word "no" is a negative particle used to express denial or refusal, while "don't know" is a phrase indicating a lack of knowledge or information about something. In this case, the contraction "dunno" combines these two elements into one word. It is important to recognize that language can be fluid and flexible, allowing for the creation of new words and expressions through processes like contraction. As such, while "dunno" may not be considered a standard term in formal writing or speech, it is commonly used in informal communication to convey uncertainty or lack of understanding. Overall, the presence of the word "no" within "dunno" adds to its meaning and serves as a reminder that language can evolve in unexpected ways.

- AI generated response

[compact holly]

it's probably a common thing with a ton of google results

[sterile crest]

Tf lol

[timber mirage]

Problem is that you did it in a certain way, and I have no idea how I should do it so it would match the conventions that you used

[sterile crest]

He didn't do anything in a certain way

[timber mirage]

There is nothing about auto updating the depdencies

[compact holly]

it's all pretty standard stuff, that every project uses

[sterile crest]

There is in a .yaml file

[timber mirage]

Well, I think ?

[compact holly]

basically 'getting stsrted guide'

[sterile crest]

The .yaml / .yml is how you create them

[timber mirage]

docker-publish.yaml
docker-verify.yaml
pre-commit.yaml
releases.yaml

[sterile crest]

Yess

[timber mirage]

there are those 4 files

[sterile crest]

Those are GHAs

[timber mirage]

There is also dependabot.yml that I have no idea what this is

[sterile crest]

Dependabot is provided by GitHub

Via a plugin

So it's not the best to reference

[stray wren]

TJ-Bot

Whats dependabot in 2 sentences

Dependabot is a GitHub bot that automatically creates pull requests to update dependencies in a repository to their latest versions. It helps keep projects secure and up-to-date by regularly scanning for outdated dependencies and suggesting updates.

- AI generated response

[sterile crest]

Look use the pre-commit as your base

[compact holly]

you can see it in action in tj-bot

[timber mirage]

Ah so it's the one I need

But huh

version: 2
updates:
  - package-ecosystem: "gradle"
    directory: "/"
    schedule:
      interval: "weekly"

nearly empty

[stray wren]

TJ-Bot

Create a GHA that can raise a PR that changes a line in my build.gradle with an output from a bash script

To create a GitHub Action (GHA) that can raise a Pull Request (PR) to change a line in your build.gradle file with an output from a bash script, follow these steps:

1. Create a new GitHub repository for your project if you haven't already done so.

2. In the root of your repository, create a new directory called .github/workflows.

3. Inside this directory, create a new YAML file (e.g., update-build-gradle.yml) and add the following code:

name: Update build.gradle

on:
  push:
    branches:
      - main

jobs:
  update_build_gradle:
    runs-on: ubuntu-latest

    steps:
      - name: Checkout code
        uses: actions/checkout@v2

      - name: Run bash script
        run: |
          OUTPUT=$(bash script.sh)
          sed -i "s/old_line/$OUTPUT/" build.gradle

      - name: Commit changes
        run: |
          git config --global user.email "[email protected]"
          git config --global user.name "Your Name"
          git add build.gradle
          git commit -m "Update build.gradle"
          git push

4. Create a bash script (e.g., script.sh) in the root of your repository to generate the output for the new line in build.gradle.

5. Commit and push these changes to your repository.

6. Once the workflow is triggered by pushing to the main branch, it will run the bash script, update the build.gradle file, commit the changes, and push them back to the repository as a PR.

This GHA will automate the process of updating your build.gradle file with an output from a bash script and raise a PR for further review and approval.

- AI generated response

[compact holly]

yeah, like i said, it's default/standard stuff

[sterile crest]

There done

And instead of push to main make it a weekly schedule

[stray wren]

TJ-Bot

How can i update java version with dependabot

To update the Java version in your project using Dependabot, follow these steps:

1. Check the current Java version in your project by running java -version in your terminal.
2. Make sure Dependabot is enabled in your repository. If it's not, you can enable it by adding a configuration file (dependabot.yml) to the root of your repository.
3. Dependabot will regularly scan your project for outdated dependencies, including the Java version. When it identifies an outdated Java version, it will create a pull request to update it.
4. Review the pull request created by Dependabot to update the Java version. Ensure that all necessary changes are included and that there are no compatibility issues with other dependencies.
5. Merge the pull request to apply the updated Java version to your project.

It's important to note that updating the Java version may require additional changes to your codebase, as newer versions of Java may introduce new features or deprecate existing ones. Make sure to test your application thoroughly after updating the Java version to ensure that it functions as expected.

If you encounter any issues during the update process, you can seek help from our community on the Discord server or refer to Dependabot's documentation for further assistance.

- AI generated response

[timber mirage]

On top of that, it needs to check if the image is available, so can I do a http request in this ?

[sterile crest]

Yeah curl

[timber mirage]

alright

i'll check that

[sterile crest]

Make a .sh that spits out the latest version and sed it into the build gradle

[timber mirage]

thank you

[sterile crest]

It kinda sucks but it'll work

[timber mirage]

heh

a simple curl should work

I just need to check the error code

or something like this

I think

[sterile crest]

Yup

[timber mirage]

well

can I use curl in this yaml file ? or do I have to create a .sh file ?

[sterile crest]

You can curl inside the yaml, yes

[timber mirage]

one last question

is there a way I can test the action in local ? or do I have to create a test project ?

[compact holly]

that's so hacky lol

and inflexible

but you know

potentially dangerous

[sterile crest]

It's not

It's just raising a PR

That's it

If it breaks, we'll see a trash commit

[sterile crest]

Easier to create a test project ig

[sterile crest]

- name: Run stuff
  run: |
    curl <>
    sed <>

[cedar fog]

https://github.com/skills/hello-github-actions

[timber mirage]

Ah and also, if I need an account to send a http request to docker, how do I handle this .

[sterile crest]

you don't

[timber mirage]

huh

this is a problem

[sterile crest]

it's not

[timber mirage]

wdym

[sterile crest]

use the docker API

[timber mirage]

it requires an account

[sterile crest]

curl -s "https://hub.docker.com/v2/repositories/library/eclipse-temurin/tags/?page_size=100" | jq '.results[].name'

anyways have fun

[timber mirage]

hmm

Alright, I'll check that... when I'll have time

[bright lark]

Oh u mean other fixes

how I forgot

Sorry will check tomorro

[timber mirage]

is it possible in the yaml to read the build.gradle ? since I would need to read the current java version...

[sterile crest]

      - name: Checkout code
        uses: actions/checkout@v4

when you do this, it'll checkout the code and you can run any linux commands you want

if you did a ls you'll see everything

so you can just read the line from gradle however you want

or better, use the gradle properties file

[timber mirage]

Problem is, how do I parse the gradle file ?

[sterile crest]

ah wait that's a spoonfeed

hehe

use grep 🙂

[timber mirage]

You are funny

But what if there are several similar lines?

[sterile crest]

will there be multiple from.image?

match on a longer regex, from.image = 'eclipse-temurin: or match on the entire jib block and work your way down that way

[timber mirage]

Hmm

I guess it could work

[sterile crest]

this is where the properties file would have been better because it's easy to parse and use

[timber mirage]

But then what I do of it ?

I need to do version + 1

[sterile crest]

so do a grep to get the current value

then use a sed to replace it with version+1

[timber mirage]

Hmm I see

[sterile crest]

something like this:

sed -i "s/eclipse-temurin:$current_version/eclipse-temurin:$((current_version + 1))/" $GRADLE_FILE

do you want me to this for you

[timber mirage]

If you want, I'll check that regardless tomorrow

[sterile crest]

okay ill do it once you give up

[timber mirage]

Note that I will have limited internet starting Saturday, so I won't be able to push any contribution this week, only do reviews

So if I can't do it tomorrow, I will only be able to do it next weekend

[timber mirage]

Actually, can you do it please? Since I will be able to focus on other things for jshell

[sterile crest]

lool

okay i can do this in less than 15 mins

[timber mirage]

5min left 👀

[sterile crest]

I can't raise a PR

fml

I need to fork the project

https://github.com/Together-Java/JShellPlaygroundBackend/pull/57

I haven't tested the commit and create pull request

the rest works

well ig i can test it

gg tho @timber mirage

took less than 15 mins

works though

[timber mirage]

Huh

Gg ahah

[sterile crest]

wym gg i can fix this in 2 seconds

whoops messed up the PR

@timber mirage is this OK?

[timber mirage]

Wrong file, see my pr, it's in the wrapper

[sterile crest]

https://github.com/Together-Java/JShellPlaygroundBackend/pull/57

now ay

lma

why is wrapper using eclipse-temurin:22-alpine and API is using eclipse-temurin:21

?

[timber mirage]

Because the api doesn't need alpine

[sterile crest]

can we not use alpine for the wrapper?

then i can just change the path to the wrapper without needing to change the logic to handle "alpine"

[timber mirage]

Alpine is lightweight

[sterile crest]

i know

[timber mirage]

It is needed so the wrapper weights less, it's critical

Anyway, I'll go to sleep
I'll merge your pr tomorrow if I don't see any problem

[sterile crest]

you can't find problems when you couldnt' find the solution 😐

😄 😄

but the PR is all yours now

I don't wanna continue with it - it exceeded my 15 mins

30 mins is too much work 😦

okay whatever, i pushed the fix to use alpine and jshellwrapper

@compact holly check out my hacky bash!!

[quartz shell]

did u guys end up switching to wazei's aws?

[sterile crest]

No

Primarily because i think everything is now working

[bright lark]

whatsup

@timber mirage im resuming the pr review fixes

and i find that it would be better to make checkJShellWrapperImageName() public

so i can reuse it in TestsConfig class

[bright lark]

done!

would you like to take a look please

[stray wren]

System out

Version: 22.0.2+9
Vendor: Eclipse Adoptium
OS: Linux
Arch: amd64

Nopox's result

Could not find session for member Nopox

[timber mirage]

@sterile crest I reviewed your pr

@bright lark you didn't address all the comments
once you addressed them, I'll validate it

[sterile crest]

feel free to make the changes, i won't be working on it - but the solution is there for when you need it

[timber mirage]

alright

[bright lark]

are u kidding ?

i dnt see anything left

huh

u want to do that selection in the constructor DockerService or Config ?.

[timber mirage]

Dockerservice
And there is another comment iirc

[bright lark]

what is it please

(cuz im lost in those +50 reviews)

[bright lark]

I said in a comment

That's because of spotlessApply

Ig

It added an empty line at the end

[timber mirage]

ah

[bright lark]

i think tht we should set git hooks

it seems tht code wasn't formatted

[bright lark]

done

[stiff oak]

@timber mirage

I'm new to open source contributions and have been exploring the code base. From what I've gathered so far, it seems that a new container is created for each session. While I’m still familiarizing myself with the entire context, I was wondering if it might be more efficient to implement a flexible pool of containers. In this approach, code snippets could be executed in an available container, which would then be refreshed (perhaps by restarting the JVM) and returned to the pool for reuse.

I understand that there may be details I’ve missed, and I apologize if I’m overlooking anything important. I would greatly appreciate your insights on this idea.

[timber mirage]

There is a concept of sessions, where if you execute several sessions in a row, you can reuse the results of the previous sessions
So it's not possible.

[stiff oak]

What if you persist the state of those sessions after execution, to resume them when they are reused or delete them after expiration.

[timber mirage]

The state is in jshell itself, I have no control on it

[stiff oak]

Alright, but is it okay, if there was a way to persist session, that is if jshell offers a way to load up and save session in some format, or is it a dead end.

[timber mirage]

it's a dead end

unless you want to research for a way

[sterile crest]

Technically

We could spin up a single container and run multiple instances of jshell

However, you get gg'd if person crashes the container then gg to everyones sessions

[stiff oak]

I would definitely like to explore this idea further. It seems, though, that you might have some reservations about the approach. If that’s the case, I’d love to hear your thoughts on why it might not be the best fit.

[stiff oak]

But how about jsh files?

[timber mirage]

• I didn't researched how you could persist JShell, but seeing how it works like a black box, I don't have much hope
• You could just save the script and reexecute them, but executing cost a lot of time, so you are moving the problem of container amount into executin speed, which is a terrible trade
• You could do what wazei said, several sessions per container, but it's always possible for someone to crash the session, or at least right now. You could fix this by making so each session is its own process, which should solve most of the issues... but then if you have several process, it would cost a lot more ram and cpu, so it's still not good enough
• Something else ?

But note that this problem isn't currently a problem

we didn't reach a point where we had too many sessions yet

if you want to improve perfs, there is another area where you could do something : creating a container and executing the startup script costs a lot of time, so it would be great if we would have a buffer of one session, so we could just assign it, making so a /jshell eval would be much faster @stiff oak

[stiff oak]

Yeah, I agree to most points. However, for one shot operations, it could be considered because I believe not all operations necessarily need a state. But since it's not a problem currently, I could stop at it.

[stiff oak]

You mean having a set of pre-initialized containers which a startup script already executed right? Or I'm I missing smth?

[timber mirage]

Containers are deleted just after the operation in such case

[timber mirage]

yes

[stiff oak]

Alright, Please let me itemize the requirements:

1. Pre-initialize some docker containers each with it's own identifying session id.
2. For every new request for /eval we allocate a container from the buffer to the new session to make execution faster.
   This is what I've gotten so far.

However I have some questions:

1. What happens when the session expires for such containers, do you want to leave them operating.
2. Does this apply or should it apply to single eval too.

[timber mirage]

The only difference between before and now, is that instead of creating a container when a new session is needed, you instead rename the buffer container to the one matching the session, and then init a new container with startup script, etc, that's all

[stiff oak]

Alright, what happens when the session expires?

[timber mirage]

nothing different, the container is deleted

[stiff oak]

Alright, it's clear, do I create an issue on that or would you instead or is it already created?

[timber mirage]

Check if there is an issue, if not, create one, I will complete it later

[stiff oak]

Alright would do that tomorrow, it's already bedtime here.

[stiff oak]

I've created an issue:
https://github.com/Together-Java/JShellPlaygroundBackend/issues/60

However, I think you may want to consider container reuse, which is essentially the same thing as earlier, only that this time, you invalidate the jshell session so that you don't have to restart another container for another session, however a mechanism like the amount of buffer containers at a time could also be implemented, like an array list, once a threshold is reached for containers in the buffer, new containers could be created and/or deleted.

[timber mirage]

It is too risky to reuse containers because users can rce

[stiff oak]

Yeah, true. So would you rather replace it with another container?

[timber mirage]

Dead sessions should have their containers deleted

[stiff oak]

Yeah, but if all containers are firstly used and then destroyed, if someone sends another /eval request it defaults to the previous implementation, so I mean when a container is used by a session, another could be re-created and added to the buffer.

[timber mirage]

The buffer should always contain one container

[stiff oak]

Alright

[timber mirage]

@sterile crest did you take into account that the GHA could create several pr if one isn't validated fast enough ?
also I can't edit the code of the pr since it's on your repo, or is it fine if I commit on your repo ?

[timber mirage]

Any advancement ?

[sterile crest]

Nope, I didn't account for that, but I did do a 15 min speed run lol

[timber mirage]

@bright lark I merged your pr

[timber mirage]

And can I commit on your fork ?

[sterile crest]

u can c&p the entire workflow

and make a new PR

[timber mirage]

-_-

[compact holly]

waz is smart, doesn't want to get blamed when this thing inevitably blows up

[sterile crest]

ala's repo, ala's responsibility

[timber mirage]

@sterile crest can you rapidly review it please or @bright lark I wish to see it working before midnight
https://github.com/Together-Java/JShellPlaygroundBackend/pull/62

[sterile crest]

left a comment

[timber mirage]

alright, what do you think now
(I trust you for the command, I didn't tested it)

[sterile crest]

testing sec

nah it didn't work

we need to fix that bit though

oh

[timber mirage]

I pushed minors changes in the echoes

[timber mirage]

shouldnt you use --set-upstream ?

[sterile crest]

no - this only happens when the PR already exists

i.e. we didn't merge the new PR and it tried to raise a new one of the same update

[timber mirage]

ah so the if failed ?

[sterile crest]

yeah

but idk why

[timber mirage]

if gh api repos/Together-Java/JShellPlaygroundBackend/branches/$branch_name >/dev/null 2>&1

well

let's test things one by one

does gh api repos/Together-Java/JShellPlaygroundBackend/branches/$branch_name works ?

[sterile crest]

nope lel

I added echo "A PR is already open for this release" but it didn't print out

[timber mirage]

But do we really need it ?

Since it will fail anyway

[sterile crest]

up to you

seeing this kinda is ugly

but if we don't have any alerting on failed actions then sure

[timber mirage]

Yes but it should only happen if we don't merge the pr within one week

I'll add a pr for that if you want to work on it$

[sterile crest]

i dont wanna work on it

[timber mirage]

Alright

done

Review it please

[sterile crest]

well approved

[timber mirage]

@sterile crest is this normal that nothing happened https://github.com/Together-Java/JShellPlaygroundBackend/actions/workflows/update-jshell-image-version.yaml

?

[compact holly]

philosophically or?

if you think about it, for the majority of time since big bang, nothing really happened

so i would say nothing happening is quite normal

[bright lark]

hello

im think im late

[sterile crest]

Did you even look at the code before merging the PR?

It very clearly stated every Friday at midnight

But you can manually kick it off since I added workflow dispatch

[bright lark]

i dnt see the PR purpose

and it seems related to previous merged PR

the title says "update java version"

but i see in changes: "name: Auto-Update JShell Image version"

[bright lark]

happy to know

BTW i'd like to raise awareness about the experience i got from the reviews done by ala it was really helpful

it was really helpful
no no this is insufficient

i mean it needs to be written on a book

[sterile crest]

2 junior developers helping each other improve, nice

[bright lark]

another junior is kidding

[sterile crest]

It's cute

wym these are junior level problems and reviews

[bright lark]

they are not

[sterile crest]

Yeah they are

[bright lark]

not at all

i mean

some were significantly important and they implicitly bring values

[sterile crest]

Yeah important sure

[bright lark]

i cannot explain deeply rn, it's just that i wanted to say that ala is a very good java player here

that doesn't mean he's the best in the whole server ofc

im talking within the context of this channel

[sterile crest]

Yeah he's just better than you

[bright lark]

ig i'll become better than him

😀