SeAT - ⚙-development - Page 1

Sorry just got internet back and hopefully it stays back

any feedback for that one ? https://github.com/eveseat/eseye/pull/80 initially, purpose was to simplify the fetch to cache logic - but since we bump php version, I also upgrade phpunit, unit testing and switch travis workflow to github action

It won't need to be updated for the depends_on part and the depends on just tells a container to start those services before itself

ah, right, container vs service, that's the thing 🙂

why not a docker-compose.override.yml instead ?

Well I mean you could use that but it's just going override those values if they are set if they aren't they'll default to the part on :-

yml
version: "3.2"
services:
  mariadb:
    container_name: ${mariadb_name:-mariadb}
  redis:
    container_name: ${redis_name:-redis}
  traefik:
    container_name: ${traefik_name:-traefik}
  seat-web:
    container_name: ${seat-web_name:-seat-web}
  seat-worker:
    container_name: ${seat-worker_name:-seat-worker}
  seat-cron:
    container_name: ${seat-cron_name:-seat-cron}

I mean yeah it could be via that

oh no, if using override, I was thinking about putting changes inside it instead of altering the .env file

just looking for the best and more logic option

also thinking about extracting traefik outside main compose and using override capability

``${VAR:-default}`` -> value of ``VAR`` if set and non-empty, otherwise ``default``

Is it possible to get a "Doesn't Contain" option?

Isn't ``Is not`` That?

I've had issues with Scopes and the "Is not" feature actually matching correctly.

What are you going for in the end?

Is and Is Not do not appear to have the same results. I'm attempting to ensure the Guest Squad I have created is only applying to members who have authed using the Guest SSO policy I have created. As it stands, members with high sec alts are also being added to the Guest Squad.

By adding a "Doesn't Contain" scope into the Guest Squad filter, I would be able to ensure those who have signed in with my Guest SSO policy are separated from fully authed members with HS or out of corp alts.

And my members alts should be kept out of the Guest squad. (I'm hoping)

End goal is to have anyone who signs in with my Guest SSO policy added to a Squad only they are in which is heavily restricted despite what other alts they have.

I just have my guest squad set as "Are they not in X Alliance" If they aren't they are a guest

Wouldn't that apply to out of corp alts?

I mean in the end you have the guest group classified so nobody can see the list of who's in it.. and even if it did they still would have access to the discord connector that you are giving in the guest squad..

You could also just set Is not for some random scope

Scope is a list, it needs contains as operator

Is not for a random scope doesn't seem to work. I need to be able to ensure that corp members are not authenticating using the Guest SSO profile and gaining access to my Discord without being fully authed.

For the time being, I have removed a fairly un-useful scope from all of the SSO profiles and added it to the Guest SSO profile. This enables me to us the Contains "specified scope" on the Guest Squad filter which does the trick.

A good scope to do this with it publicData

Discord connector won't do anything unless they are in the ACL for the connector, If they're not in the ACL it'll let them join but your discord should be setup properly so stuff is locked behind roles.. if it's not that's on you.

Discord is setup fine. My work around seems to do the trick for the time being seeming as there isn't a "Doesn't Contain" option.

https://docs.google.com/forms/d/e/1FAIpQLScxiUuVJMiflZ2yU44pUJLVy804K0YmJTNJCYZgrXO7w2YOjw/viewform - interesting for devs imo

Maybe plugin devs too tbh

Hello @user_614098468218339348 I am going to start looking at seat-inventory and add a market source, do you know if there is a library, or plugin, that can help me to check the market sell orders for an item ? I want to look at a player owned citadel

it looks like this API endpoint will return orders https://esi.evetech.net/latest/markets/structures/{structureid]/?datasource=tranquility&page=1

yes, you basically implement a job that fetches this. Then there are two options: directly store it in an inventory source or store it in a table more optimized for orders, then copy the data over to inventory sources.

additionally, you might want to look at seat-treelib, it currently fetches public orders

also check if seat doesn't already fetch your citadel orders

if seat can fetch the citadel orders, then that's the way to go I think, I will only copy the data over to inventory sources then

such functionality exists in seat ? because it require the access to the citadel

so it need to use the auth of one character that have access to it

take a look at the eveapi package

thank you

so it doesn't look like it can pull orders from a structure, only player own orders : https://github.com/eveseat/eveapi/tree/master/src/Jobs/Market

I see you pull public orders yes here : https://github.com/recursivetree/seat-treelib/blob/master/src/Jobs/Orders.php So I basically do the same, but for a player owned citadel. I will check if the API is the same.

both are really really close APIs /markets/structures/{structureid}/ on the left /markets/{regionid}/orders/ on the right

👍

it's better if you can use the work I do @user_614098468218339348 You want me to try to add a job and migration to seat-treelib or to seat-inventory ? What would make more sense for you ?

Wuuuut > 500 days of data When in the life did they think it would be a good design option 😮 You even can't load that old for wallet transactions, common

Regarding ESI orders, maybe that would even be better if it lived in eveapi 😛

It's also a possibility but I need a bit of guidance because it's a bigger package

Only talked about the orders job

So mostly clone the existing job, update endpoint, adapt needs

The hardest part will probably be to determine if you can run the job for the structure - not sure we store structure ACL currently And we shouldn't make invalid calls against esi

https://github.com/eveseat/eveapi/blob/5.0.x/src/Jobs/Market/Character/Orders.php

a month ago you didn't want to merge order loading into seat 4. If you change your mind, I can put the PR back up :)

we don't store acls and there is no way to figure the acls out. It's basically the same as with the structure loading PR for seat 5

however, you can find characters that are likely to have access, for example by looking if they have assets or jump clones there

if you add order loading from esi, please add it to treelib, so all order loading code not from core is in treelib.

Wuw, did I ? 😮

Said SeAT 4 as I want to get rid of it and release the 5 one 😅

maybe I misunderstood it. Anyways

What is holding seat 5 back from being released?

No clue, didn't play with SeAT 😅

If there are non blocking issues, I can tag stuff anytime probably

I went through the repos, these PRs might still be relevant: https://github.com/eveseat/web/pull/635 - qu1ck/anyone else? at the same time, it's not very important https://github.com/eveseat/eseye/pull/80 - I think it can be merged https://github.com/eveseat/seat-docker/pull/21 - docker stuff, need to look at the details https://github.com/eveseat/seat-docker/pull/20 - docker stuff, need to look at the details https://github.com/eveseat/notifications/pull/60 - native discord notification. not completely done, maybe better for seat 6 or something like 5.1 Also I think @user_301981661761896449 was talking about some docker stuff that isn't in any PR yet

First is for @user_398925835794645002 Sec stuff is him

I think we also need to rebase, there are some bugfixes like the one with sde tables that aren't included yet

There will be a merge

for the arguments one i think were good. i couldnt find an obvious way to abuse it

I will put into treelib then if it's more consistent with the rest 🙂

but I need to play around a bit before because it's all new for me currently, I need to build experience

For me the only current blockers are updating seat-docker with changes to rerduce boot up time by not needing to run the permissions fix. This goes hand in hand with not running as root within the container, but is not as functional as I thought I had it. The only other thing I can thing of (and probably not needed for an RC) would be an update of the analytics code. This I have started, moving from UA over to GA4. But apologies not been able to get this across yet

Is there any documentation on how to create a skin for seat? Do I need to manually do everything or is there something like a generator?

Currently no, I had worked on a customized dark version a few years ago but dropped it. The more difficult part is SeAT does not dynamically build the skin list.

the second thing is not an issue. I can create a PR to change it.

it is bootstrap based, I don't know what kind of doc you want else of this 🙂

css

I mean, there are two files in the codebase that reference "Jet", pretty easy to make a dark mode skin and add it to the hardcoded php settings file.

User > Profile > SeAT Skin

I imagine a global option could be added to Settings that populates from a directory of stylesheets and the option in User could be a Light/Dark mode toggle. Granted if a theme doesn't include a dark mode version it should be disabled.

If it didn't require me to learn PHP and probably multiple new libs Id give it a crack myself :/

I think I draft something for dynamic settings and their default value while pocing for 6.x, but not sure it's been pushed

Related to skins, we might mimic what WordPress is doing in order to load skins dynamically based on css header block

css
/**
 * @name Jet
 * @author Warlof Tutsimo <warlof@eveseat.net>
 */

Then, required the css file is published against a specific directory (as actual affaict), and simply use a provider to copy the css file from the package to the directory when installed

Either this or use the provider to seed config with elements

I think it's easier to implement just skins like we have it right now. If you want dark mode, you select a dark mode skin, if you want light mode, you select a light mode skin.

there seems to be some interest in more skins, should I target seat 5 or 4?

https://github.com/eveseat/web/pull/647

Would need an update into services package aswel

For what?

Just noticed it. I changed Profile locally but didn’t realize it is in services

I want to create a job to get StructureMarket orders. It will target /v1/markets/structures/{structure_id}/ and it is an auth one. So I am extending AbstractAuthCharacterJob. I see that Seat use a Bus to fire all the jobs regarding characters for Example (Seat\Eveapi\Bus\Character) Is that the right way to get a good token for my job to run against ? $token = RefreshToken::find('character_id'); Do I understand right if I say that Seat\Eveapi\Commands\Seat\Buckets is in charge of refreshing api tokens for all characters registered to seat so I can use RefreshToken safely ?

yes, that's generally the way. I'm not sure it belongs into Bus tho, we don't need to fetch it with every character every time we try to pull data. Also you need to check what happens if ACLs don't allow you to access a particular structure. You don't need to worry about refreshing tokens as long as you use a job like AbstractAuthCharacterJob. Just pick a token and schedule the job.

ok because I get an error right now

aaaaand the problem was somewhere else as usual haha

now I get orders that's fine 🙂

nice

https://github.com/eveseat/web/pull/647 can be merged, the PR to services is done

Someone asked me to improve the moon reporter tool. What is a ore moon? And shouldn't Gas be R4?

When was the last time the code for that was touched? On 3/30/2020 CCP removed asteroid ores from moons so if the code is from before that time the ORE bit might be leftover, and yeah I think Gas would be R4.

@user_301981661761896449 Can you take a look at the other Citadel PR as well when you merge?

https://github.com/eveseat/eveapi/pull/400

Would pull me off having to run a franken-merge Dev install in my live environment 🙂

I'll have to do that tomorrow now sorry. And I want to check the ci set up again as something went wrong just then with the auto docker builds and I had to do it myself.

All good

Initial reaction, as you multiply by a decimal does it make the const a float? If so it probably makes no difference, but just something I tend to avoid normally as in my other endeavours floats slow things down a lot.

But that's not related to me merging it. Just my weary overthinking brain

reminds me of programming old PLCs 🙂

"DO NOT USE REAL, save the analog values as INT with a 10x multiplier out of the scaler"

"remember to divide by 10 on the HMI"

🥳

We could just convert it to a int after the multiplication. it doesn't really matter if it rounds 52.0001 to 52 or 53, both values are good enough

that works 🙂

I tried to specify the const as an int but got red squigglies and was not certain of the syntax.

(Really rusty in PHP still 🤣 )

(int)($base*0.65)does it I think

but also, const in PHP sometimes doesn't allow things that could be const

This was where I felt a little off the path 😅

why that multiplier?

because 65% of another value

Casting is not supported for a constant. "Constant expression contains invalid operations" We could do const RATE_LIMIT = parent::RATE_LIMIT - 30 instead which would avoid casting it. Or just plain const RATE_LIMIT = 50.

Haha I didn't mean to change anytjing sorry!!! It was just my brain on a late night

@user_301981661761896449 @user_76781846418956288 fix so that new instances work again: https://github.com/eveseat/eveapi/pull/401

I ran into this today as well 👆 I... uhmmm.... accidentally mixed up two WSL2 commands and wiped my whole dev environment 😶 So when I did a fresh install I got the same problem that others has been reporting.

welcome in the club. I usually end up wiping my development install's database because I managed to do stupid things on average around once every 3 months

Yeahh.... This was the entire Linux dist 🤣 With like 4 active pet projects needed to be restored 😫

That's when you thank your lucky star that you are using Docker and have automated scripts with DB init and everything

Available, thanks for the diag

I've been keeping an eye on the Redis fork landscape after the licence change. The Valkey https://github.com/valkey-io/valkey project is the fork that has been adopted by Linux Foundation and has one of the 5 previous maintainers of Redis helming the initiative. It's already backed by Amazon Web Services (AWS), Google Cloud, Oracle, Ericsson and Snap Inc. Microsoft recently released their own Redis client compliant Cache server which they will be focusing on going forward https://github.com/microsoft/garnet Worth to note is that Snap Inc (Snapchat) who has been backing and maintaining one of the other big contenders KeyDB https://docs.keydb.dev/ which supports multi threading for Redis is also one of the backers of Valkey. This might be an isssue in the long term because why would they put full max effort into two competing similar projects.

reminds me we should do some performance profiling of both 😄

In the Market Prices table, what're the average and adjusted prices?

Well, I guess they're the CCP provided average and adjusted prices, but what even are those? Average over how long, adjusted how?

nvm got answer elsewhere, tl:dr its a sekret

pretty much, they are just two values somehow related to prices

I'd recommend to use the price provider system if you are writing a plugin, or the sell and buy orders columns

I'm doing something far more terrible than that.

Quick question, to get the latest eveapi on my server, do I have to wait for an update on the docker image ? Is there a way to update it ?

In theory yes it is possible to update it, but the correct approach is to wait for the docker image...

Its coming shortly

Oh okay no problem I can wait

I was just surprised to still see the previous one after rebuilding my container, but no issue i'll wait ^^

(I thought it was updated like the plugins are)

hehe yeah it wont be there yet as I only just tagged a new eveapi, and have yet to build the container as I am putting a few other things in first

And no, core will not by default update on down/up

That is so that any give core image is at least semi repeatable

Yes I see it makes sense

docker building now.. So about an hour away from completion

Great !

So if i wanted to edit how the embed displays killmails in notifications would i need to set up a dev container or can i just edit it and save ?

Should do a dev install otherwise your changes will be lost each time you reboot.

Alternatively, you can make your changes and submit them as a PR to core and sharing changes across the world 👌

I would definitely think about it if I can contribute anything useful. For now I'm just looking at cosmetic changes

even cosmetic improvements are appreciated

You can also add the custom-layout.css file. While this is not ideal you can change A LOT of stuff this way. This is how mine looks now

just a couple of screens as an example...

Wow I love that look. Did you edit the .css or is that the default custom -layout

yes

Haven't finished up everything yet as as started a few days back

here's my custom styling - copy do not use the remote url in you app please

https://seat.eve-ro.com/custom-layout.css

Thank you I will check it out

The advantage is that if they change something on the logic (blades) nothing will be impacted, some styling may be broken but your app will work just fine. You can always remove the custom styling and will fallback to original

Here’s another EVE project I played with a few years back… it’s still online lol http://www.eveshipsdb.com/

The webgl / canvas is a bit buggy as this is a React project but will regenerte the ship if you refresh 😉 eg: http://www.eveshipsdb.com/ship/22448

How would you load it if someone is using a docker container?

https://eveseat.github.io/docs/styling/

I just folowed these steps

Thank god for you

seriously

have my children

Bare in mind that the file is work in progress... I am working on it and will take a while as the styling is all over the place currently, there's a ton of inconsistencies that are quite difficult to sort out without breaking something elsewhere.

its so good though

It will get better 😉

just popp'd it on my v5 test instance

made my entire body relax its so good lol

I hated the original skin so I built my own hehe

Just check the file once in a while as I am updating it whenever I got the time

Making changes to it daily, adding missing elements, etc so bare with me hehe

You might want to look into publishing it as a plugin providing it as a skin. I think I did the work so that plugins can publish skins, but then I never wrote the skin.

I might look into this later on…

Technically this is not a skin tho 😁 sorta

At least at the moment I mean

It would be nice to have a way to easily override the blades tho… there’s so much that can be improved… data is all in there - you can do whatever you want on the FE

There's a ton of data that's in there and not used at all...

The skins are pretty much a css file that is included after the default stylesheet. So it is not at all different from a style override

Sorry, I'm new here and I'm having trouble updating the seat.

#channel_821361546608508938

front-1 | front-1 | INFO Routes cached successfully. front-1 | front-1 | front-1 | INFO Running migrations. front-1 | front-1 | 2020_03_26_110237_convert_to_seat_400 ............................. 1ms FAIL front-1 | front-1 | In Connection.php line 829: front-1 | front-1 | SQLSTATE[42S22]: Column not found: 1054 Unknown column 'group_id' in 'field front-1 | list' (Connection: mysql, SQL: select distinct groupid from seatconne front-1 | ctorusers order by groupid asc limit 1000 offset 0) front-1 | front-1 | front-1 | In Connection.php line 423: front-1 | front-1 | SQLSTATE[42S22]: Column not found: 1054 Unknown column 'group_id' in 'field front-1 | list' front-1 | front-1 |

thanks

I installed SeAT 5 with keydb replacing redis yesterday. Seems to work really well, but it DOES have the same issues redis has (it can spiral out of control when jobs fail). I didn't have to make any config changes, it just worked™️ out of the box

with redis changing the licensing as of late, it might be worth looking at alternatives

it's also faster

by a little bit

(from https://github.com/Snapchat/KeyDB )

I tend to agree, but also with the little time that is currently spent on seat development, I'm not sure it has that high of a priority. Additionally, the license changes doesn't affect us. Also, let's wait a bit to see which fork will survive and become the new de-facto standard

indeed

I will keep things on keydb on my install for the time being

it's bare metal

4000+ tokens

https://devops.com/valkey-is-rapidly-overtaking-redis/

looks like there are a few other alternatives, too

@user_169137647225143296 Have you seen the same fault on the assets job? I ran it briefly on a new test instance and didnt observe a failure, but not sure if there is a condition I am not meeting.

I have a fix regardless that will no longer use the mapping to modify the timestamp, but its embedded in a larger change that I need to be at my home pc to push ( 18h minimum)

If it didn’t run at least an hour after an asset job entered the queue you might have inadvertently missed it, it sits in pending and retries for an hour before the job itself fails, but the laravel log shows the error immediately.

@user_301981661761896449 I submitted a PR with a fix for the assets job

Hey mate! That looks good to me. Could you update the PR though to make the same change to the Corp assets job?

Yeah my error... I tested the initial pull. Mind blank because this change was related to unchanged items.

Have you looked into Kvrocks? Since it does not rely on the memory but the DB this could be the better angle?

nope

I just tried a slot-in replacement for shits and giggles since it's multithreaded

not like SeAT really pushes redis hard

swapped back to redis because of some citadel errors ... but those remained with redis

so that'll be fun

I suspect that's a legitimate bug

you can use kvrocks as drop-in replacement as well, but as said it does nto use the memory but has a persistent volume. This is actually what borked redis with a corrupt memory dump in the past

yeah... I have enough memory, I want to reduce disk trashing

Regarding @user_98470336932093952 issue in #channel_821361546608508938 : https://discord.com/channels/821361165791133716/821361546608508938/1247297270861402163 It seem like it has recently been introduced by https://github.com/eveseat/docs/pull/168 It has several changes and additions referencing a .env.example file in https://github.com/eveseat/seat-docker, but that file doesn't seem to exist there. Any context on this @user_301981661761896449 @user_164076488294006785

.env.example is not on the master branch, it's on the 5.X branch

but the docs reference master

also, seat-docker is the only module where the seat 5 code is in master

seat-docker:5.0.x is outdated

the pr should have been on master

By chance, do you have an updated version?

Those is the latest version… haven’t done much on it, just go to that url and refresh

I will move the details across tonight

https://developers.eveonline.com/blog/article/updates-for-equinox-developers-edition

will need to keep an eye on fuzzworks sde dumps, and if it changes, adapt.

the good thing about fuzzworks dumps is that he tries to keep the stable even if the sde changes. Already now, the tables don't map 1:1 to the sde and some data, like repackaged volumes of ships, is introduced from other sources

@user_131158012831203328 can I get the horizon page of one of your failed citadel jobs? I do have a new theory about what could be causing them

I uh

I'll have a look

I did an update, added a plugin from crypta with citadels etc, so I blew the cache away

got some

@Wibla can I get the horizon page of

is it possible change the display date format in the table wallet->journal??

thats iso date format

Does that mean no?

Currently not changeable

I'd rather see the date and time than xx minutes ago

@user_131158012831203328 @user_155822088693743616 Can you guys test this PR? It should reduce the citadel load a bit: https://github.com/eveseat/eveapi/pull/407

interesting

ok so I'm going to ask a really stupid question here

the citadel jobs look up structure names?

or am I missing something?

Yes, that is its main purpose

oh boy

you don't need to check that per charid yeah

haha

taking a backup now, then I'll test

testing

still getting a ton of citadel errors hm

replaced the old files with new in vendor/eveseat/eveapi/src/Jobs/Universe/Structures

and restarted all the things

i think ive added it

erm, getting 404 page not found

had to remove it, was causing 404?

sry each attempt to add it causes 404

It won't make the errors go away. The citadel issue with ESI means it will always be there. This should just decrease the rate

The reason why multiple ids were (and will be, just spaced out) checked is that we have know way of knowing who can access that citadel information.

It doesn't get added as a plugin is why, the code would need to be added as a dev install for docker

i aint at that level yet. adding stuff to env is about my limit

seems to work now though... errors are steadily dropping

Is there a reason we only schedule one director per corporation?: https://github.com/eveseat/eveapi/blob/5.0.x/src/Commands/Seat/Buckets/Update.php#L111

What is the best approach to add translation to a html text fragment containing links? For example:

html
Edit and create roles on the <a href="{{ route('seatcore::configuration.access.roles') }}">access management</a> page and automate role assignment using <a href="{{ route('seatcore::squads.index') }}">squads</a>.

I could split it into multiple trans() calls for each text segment, but that is difficult to understand when looking at the translation file. Also in my experience, other languages sometimes reorder things in sentences and then this structure is too inflexible. The other option is to have html in the translation and put it into the browser unescaped, but I don't like that either.

Could you have placeholders in the trans call for the links like and replace them again afterwards?

probably something like that if it there is nothing from laravel

so for structures, I'd probably do something like ... "ok this character works -> put it as the first on the list for refreshing a structure"

in a database table, not in redis 😛

Redis vs db aside for blacklist. Once a structure has been resolved at all, it is not refreshed.

is there any way we can get the assets fails to try twice then wait a day? instead of trying over and over? 95% of our fails atm are assets failing

Failed jobs or logs you are looking at?

jobs

But no there is no way to do that atm

There is some logic in the latest update that will reduce the time spent on failing jobs, but it won't reduce the number of jobs that fail in theory

been 22h since i did the update and some are already on attempt 20

feels like its adding to server overhead without acheving anything

Attempt 20 can be deceptive though... The attempt count gets incremented for any reason if the job is put back on the queue.

Ie if the error rate limiter kicks in and delays the job, that counts as an attempt even though the job was never run

Guess who is making about 15000 api requests every morning to check the lowest price of a bpc in jita 👀

that makes an in-memory cache a bit more valuable heh

ahahahah

yeeeah

so fucking dumb

"oh yeah we had to scrap the XML API because of load on the monolith"

what about... no you didn't

you could have put into place the same abstraction layer with XML

you just wanted to do something fancy and modern and aaaaaaaaaaaaaaaaaaaaaa get fucked

(there were other issues with the old XML API, don't get me wrong, but it was trivially simple to work with)

ok I guess that became a monologue / rant

sorry about that 😄

I understand how esi is better for the monolith than xml was. I understand some of the practices they attempted to institute. And most of all I am vaguely aware of intracies of the monolith that dictate some of the design of esi that are often seen as undesirable. But I think the biggest thing that plagues esi is stagnation.

I wonder how much they pay in egress fees

to AWS or whoever is hosting the ESI shit

@user_301981661761896449 just a thought, but if you have any ideas for database performance profiling that you want to test out on a 4300+ token install, I'd be happy to help

could be there are some low-hanging fruit there that is easier to spot on a busy install

Esi was a cool thing when there was a team working on it Aka snowedin and others

Last time I play with it, it was aws stuff

At the very least the load balancer is AWS as that's the layer that strips off the user agent.

9 days, 5.5TBW on the ZFS SLOG drive

💀

Any chance the https://github.com/eveseat/notifications/pull/101 and https://github.com/eveseat/notifications/pull/102 could get merged? I noticed today that StructureUnderAttack which is also used in the Skyhook PR failed if the agressor was not in an Alliance

@user_301981661761896449

*Noticed and fixed in my PR

Feature Request: BPO, BPC library and/or request system

That sounds more like a plugin than a core feature

Very likely true

It's also something NO ONE is doing as far as I can tell and there is a lot of need for something like that.

Depending on how important it is to you, you can complain about the lack of a plugin, or take it into your hands and get a plugin made.

You can: 1) learn to program yourself 2) convince someone in your corp/alliance to do it 3) convince someone from the seat community to do it 4) hire someone to do it and pay him, be it ISK or $. IDK if this works and how expensive it would be but no plugin programs itself from nothing

@user_301981661761896449 Is there a reason we don't have all migrations when running the tests?

@user_76781846418956288 I'm trying to write a new connector plugin for seat-connector and I don't know how to solve this problem.

What were you trying to do when this happens?

Also, the composer.json doesn’t matter for starting up the docker seat stack

I have successfully let Seat load my plugin, but is there no way to hot update SeatDocker?

@user_614098468218339348 why my i18n not working...

https://eveseat.github.io/docs/developer_guides/developer_installation/

you need the locale middleware on your routes I think. https://github.com/recursivetree/seat-billing/blob/main/src/Http/routes.php#L6

ok thank

wait it i already use it..

you might need to php artisan config:cache or something like that

and route:clear

same ...

Whatever, I've already implemented the registration logic, so I'll just leave it at that for today.

But thank you anyway

Did you manage to get it working? If it is on github i can also take a detailed look

At least the controller looks fine, but there is still a problem with i18n. I will upload it to github. Can you help me solve the i18n problem?

@user_76781846418956288

Are you there?

https://tenor.com/view/spray-bottle-cat-spray-bottle-spray-bottle-meme-loop-gif-25594440

so connector_type mast be lowercase?

Because when I use uppercase characters, I can't find the Driver Class

https://github.com/FeiBam/seat-qq-connector

Can you help me with the i18n issue? Here is the github address

Seems to work for me

have you tried to restart the docker stack?

I have fixed the bug yesterday, I just released the first version and submitted a pull request to seat doc

great!

@user_301981661761896449 Any ETA on publishing a new docker release with the stuff merged yesterday? Or should I run dev-install on them in the meantime?

Tonight

Awesome! Thanks!

Some bugs..

hmmmm

Found 1 security vulnerability advisory affecting 1 package:
+-------------------+----------------------------------------------------------------------------------+
| Package           | twbs/bootstrap                                                                   |
| Severity          | medium                                                                           |
| CVE               | CVE-2024-6531                                                                    |
| Title             | Bootstrap Cross-Site Scripting (XSS) vulnerability                               |
| URL               | https://github.com/advisories/GHSA-vc8w-jr9v-vj7f                                |
| Affected versions | >=4.0.0,<=4.6.2                                                                  |
| Reported at       | 2024-07-11T18:31:14+00:00                                                        |
+-------------------+----------------------------------------------------------------------------------+

Found 4 abandoned packages:
+-----------------------------------------+----------------------------------------------------------------------------------+
| Abandoned Package                       | Suggested Replacement                                                            |
+-----------------------------------------+----------------------------------------------------------------------------------+
| web-token/jwt-checker                   | web-token/jwt-library                                                            |
| web-token/jwt-signature-algorithm-ecdsa | web-token/jwt-library                                                            |
| web-token/jwt-signature-algorithm-hmac  | web-token/jwt-library                                                            |
| web-token/jwt-signature-algorithm-rsa   | web-token/jwt-library                                                            |
+-----------------------------------------+----------------------------------------------------------------------------------+

Does not affect seat core at this time

Yeah they have been that way for a while. Doesn't stop them functioning for us

nope, but abandoned packages are a possible security risk

Only in that security vulnerabilities are not fixed if found. Until such time they are no greater risk than a current package. I'm not saying we shouldn't move away, and I plan to. But it's not something that is simple swap out either

Actually that's not the package I was thinking of... Its just been renamed I think..

Yeah 🙂

Kind of.. More to it than a simple change

as long as it's not forgotten, as they say 🙂

It's on the list... But so is a lot of things with an absence of time to do them

How would yall feel about a PR that added iulixxi's skin?

what is iulixxi's skin?

but in general, yes

https://discord.com/channels/821361165791133716/821361505735278592/1235116990432284702

been running his 2.0 version for a while and havent seen any broken stuff

https://devdocs.io/ y'all seen this?

what would the id value be?

This is the bit you didn't like above, probably a map key returned from eseye when eveapi configures that cache within eseye. So flow would be like When eseye singleton is inited by eveapi For each cache eveapi wants, tell eseye to create that cache Eseye creates the cache the same way it does now but stores it in a map, returning the key to the map. Eveapi stores that key for later use. When making a withCache call pass the key to select the cache. To handle the case of no key given, a default is setup at the beginning. Invalid keys result in exceptions.

That's my current thinking anyway...

But what has me thinking also is about that tempCache internal variable that would be needed withCache to be able have a check in invoke to use that cache. As I'm not sure how the singleton could allow that variable to lead across and smear cache changes across calls.

So perhaps instead of withCache builder method, as you said earlier maybe an alternate invoke method that takes an extra argument to choose the cache

But yes lot of testing required definitely

From your first option of having multiple eseye objects, is that something that would be expensive to have do you think? I'm not sure on the size of the object.

File cache is one thing, can I wish for a look at how the Redis cache keys are handled with TTL and such so that I don't have to run a cronjob every 24H to clear the cache?

crontab -l -u www-data
* * * * * php /var/www/seat/artisan schedule:run >> /dev/null 2>&1
55 23 * * * cd /var/www/seat && php artisan cache:clear  >> /dev/null 2>&1

currently have this abomination in my crontab

I have looked at the ttl.... It's using the etag expiry, so will stabilise on roughly the timebase of a full bucket cycle (plus a bit of overhead)

And these changes will affect all caches not just file

Which to me appears apparent that your redis cache is stabilising before you clear it https://discord.com/channels/821361165791133716/821361546608508938/1283015248923590758

somewhat, I've set it to evict keys based on volatlile-lru

I'll suspend the crontab cache clear tonight and see how it fares over >24 hours

Now is also probably a bad time to test.. As redis is eaten up by failed jobs.... And esi is not behaving.

very true

reverting that change

😄

used to sit around 300k

now it's ... oof 🙂

I dont think its expensive to be honest. Especially because the object only really holds references to other things. In fact, I ask about the singleton because maybe its time to init per worker? At least that is something I think would be an interesting test. It also means you dont have to hack anything eseye and just config the cache the worker wants.

Yeah its a tricky, and complex design I think. Wonder if per worker init is not the way go, assuming my init time/load assumptions are true ofc.

https://i.imgur.com/cukoItF.png hello there zen5

is it possible to pass the token myself to seat ? for instance if i wanted a front page for my alliance. with differnt eve applicatons. but only want them to login to eve once.

So the refresh token can change when ever you fetch a new access token, it happens periodically and unpredictably (afaik) but that makes multiple services which use the refresh tokens pretty hard to do

also, there is not a way to do it by default with seat, but you could do some voodoo combining the user api end point to createusers and then artisinally inserting their tokens into the db...but you are going to have sketchie issues

ok. doesnt sound like its worth it to do.

You'd also have to create some sort of central service which syncs the refresh tokens across all your apps

I think you'll find nothing that needs the access token will play nice with that sort of auth

but for things that dont need any scopes, its pretty easy to do

(not a seat thing though)

they will all need scopes. I will just make them login twice lol

whats the other app?

pathfinder

oh yea, super low bar then 😛

yep

You should not be sharing tokens between applications

This is a core tenant of the design of ESI and should be clear in the third party developer agreement

theres nothing specifically about a single developer using tokens in multiple disclosed usages. Just undisclosed stuff. if you created a page that said "login to seat and pathfinder" that would satisfy the requirements just fine. It isnt how ever a clean cut easy solution from a technical perspective

Yeah

don't be the guy that ccp says we all are

Did anybody receive a structure lost armor or structure lost shield notification recently and is willing to dig it out from the DB? I do need one to do some testing, but I don't own any citadels to test with

I'll dig Sorted 👍

So I've been thinking about doing a personal project for eve. Just for fun. I have no idea on how to handle login/token information. Any recommendations or websites that you guys know of that would help would be appreciated

Well SeAT is one example! Though the code for various parts of the flow are scattered across eseye and web packages.

Have you had a look at the eve docs?

https://docs.esi.evetech.net/

I've only had time to glance . I'll dive deeper into it. Any chance you know where I can look in seat for how to handle the tokens ?

Which part of handle the tokens is the question? But best place to start would probably be eseye (handles using them once you have a refresh token)

Awesome . Thank you

I'm looking for how to get them to login and return the token so I can pass that to the esi. And how to do it securely

OK so for that side of it that is handled in the Web repo for seat

Have a look in the SsoController

Thank you again. For now it's just for fun. But if I create anything useful I'll be sure to post

Enjoy it!

Also keep in mind seat is built around laravel which does a lot of work for us. And any different languages will handle things slightly differently, but the basics are the same being oauth

👍

Is python a good choice ?

It's definitely not a bad choice in my mind.

If you want to have a look at a good web app in python and django, check out AllianceAuth

Hi

https://allianceauth.readthedocs.io/en/v4.4.2/development/index.html

Thank you. I will check it out

@user_301981661761896449 that's a very well put together roadmap document

Was an excuse to practice LaTeX

it's better put together than some stuff I've seen at work...

is there an easy way to generate a correct redirect url ?

when creating a redirect url i dont understand what to put as the state parameter. is there a list somehwere ?

state=<something-unique>

https://docs.esi.evetech.net/docs/sso/web_based_sso_flow.html contains some guidance and a link as a starting point for your own research

yea ive been reading this. im just having a hrdtime understadning the state parameter

It's to prevent Cross-site request forgery (CSRF) https://auth0.com/docs/secure/attack-protection/state-parameters https://en.wikipedia.org/wiki/Cross-site_request_forgery Is there any specific part you have trouble understanding?

Generically... Think of the state parameter as a way that you can tell you sent this person to eveo to auth. Often services will drop it in a secure cookie before redirecting to eveo.

This prevents some one pretending they are you and getting folk to click links on their pages that do things on your site

Thank you guys. I understand it way better now. I wrote a script to generate the oauth url and randomly generate the state param. and it worked finally. 🙂 now on to getting the token and such

What's the prefered way to make feature suggestions that isn't just gonna annoy folk

generally just suggest it to the relevant people. If you don't want to annoy people, just stay realistic with your expectations of it being actually implemented. If you've understood that we aren't getting paid to do this, you should be mostly fine

I would like a way for a CEO to be able to associate unregistered on seat Toons with main accounts. Use case nice: some times you have folk in your corp that go afk for long periods of time, but you dont want to kick them. So them, and their alts would be nice to group them together in a reportable function. It Also allows seat admins to chase folk to get toons registered with out knowing who the mains are. Use case CI: Having his tracked anywhere, would allow plugins to prevent or alert when an 'alt' signs up for services as a Main

To me, that sounds more like something for a plugin. Or why should it be in the core?

I don't know either direction tbh. Alt management is something we speed a ton of time on

trying to think of things that might make neat features

hmmm, that'd be interesting

Was the issue fixed where a toon would disapear from the list of linked toon for a user if the token got invalidated? I know we discussed it but unsure if it was fixed.

Because to me thats related to the CI use case above. When the token got invalidated and you checked that character it was no longer showing the other linked characters that still had tokens. It was just "floating" on its own

I think this fixed what I am talking about above 👍 https://github.com/eveseat/web/pull/657

o/ I have a question about plugin development and the SDE. How does it works ? I saw in other projects you can define a Config/sde.tables.php that is loaded then in the plugin init in seat.sde.tables, but how does it works for the model behind ? Does it exists already but is just not populated ?

So the config allows you to pull in the database table. But yes, you will need to create your own Model pointing to that data.

oh okay ! thank you 🙂

hmmm, did CCP add notifications for merc dens to ESI? ... they did :v

Yes I’ve been working with CCP on this

Refer to glue

Hello All. While troubleshooting a scheduling issue, I just noticed there is NO reference in the setup/readme for what the Cron job schedules should be. In fact, it doesn't refernce them at all. https://github.com/pyTonicis/seat-corp-mining-tax

looking at it, it seeds the cron schedule when tables get migrated tax update every hour tax:update 2 * * * * create tax contracts every month at 2.nd day tax:generator 1 1 2 * * contract status update every hour tax:contracts 2 * * * *

Yes that Using migrations to add to the schedule is generally a bad idea. If required commands are deleted, they are gone and cannot easily be readded. It can also lead to issues when the schema of the schedule table changes. The recommended way is to use schedule seeders: https://eveseat.github.io/docs/developer_guides/package_development/#jobs-schedules

The website to create ESI applications has been updated. I guess we should check if out installation guides still work: https://developers.eveonline.com/ https://developers.eveonline.com/blog/and-once-again-welcome-to-developers-eveonline-com

Hello ! I'm having a dev question. In my industry plugin I'm looking to compute the rig effect on the production (increase material efficiency for example). So I'm looking at dogma and some documentation about it. Mainly https://docs.esi.evetech.net/docs/dogma.html I have found the attributes I'm interested in, like this one https://esi.evetech.net/v1/dogma/attributes/2538 which is the attribute modifying the material efficiency for ship modules etc. So far so good. I've looked at the dogma effects and can find the ones associated with my rig. I look at the modifierInfo and I see there what kind of operation I need to compute, like a ItemModifier that modify an item. I've been testing data on the Thukker rig, and here is the rig effect modifierInfo :

- domain: structureID
  func: ItemModifier
  modifiedAttributeID: 2557
  modifyingAttributeID: 2594
  operation: 6 

Here the modified attribute is https://esi.evetech.net/v1/dogma/attributes/2557 so it's one of my attribute I'm looking for, and the modifying attribute is https://esi.evetech.net/v1/dogma/attributes/2594 which is the Material Reduction bonus. For the type it has a valueFloat of -2. So I have found my -2% Material Efficiency rig for ship modules etc. So far so good. I also have this other effect :

structureEngineeringRigSecurityModification    "
- domain: itemID
  func: ItemModifier
  modifiedAttributeID: 2595
  modifyingAttributeID: 2358
  operation: 4
- domain: itemID
  func: ItemModifier
  modifiedAttributeID: 2594
  modifyingAttributeID: 2358
  operation: 4
- domain: itemID
  func: ItemModifier
  modifiedAttributeID: 2593
  modifyingAttributeID: 2358
  operation: 4
- domain: itemID
  func: ItemModifier
  modifiedAttributeID: 2653
  modifyingAttributeID: 2358
  operation: 4
"

This one is about applying the security level modifier, so in lowSec it's 1.9 for example. Here, the second item in the array is again the material efficiency being modified. attribute id 2594 is modified, but the modifying attribute is https://esi.evetech.net/v1/dogma/attributes/2358 and this one has no description or displayed name but is from the "bonus" attribute category. I cant seem to find how I link the attribute https://esi.evetech.net/v1/dogma/attributes/2355 (High Sec) https://esi.evetech.net/v1/dogma/attributes/2356 (Low Sec) https://esi.evetech.net/v1/dogma/attributes/2357 (NS & WH) to this 2358 attribute and then apply it. I know the result should be for a rig deployed in low sec -2 * 1.9, so clearly I need to multiply the LowSec attribute to the Material Reduction bonus, but using the effects and the attributes and I dont see how I do this operation + I cant seem to find where I can find a list of all operation with what they do (the operation in modifierInfo) Thanks for the help 🙂

My guess is maybe I should have 2358 as a constant in my code, meaning "compute security system operation", and then look for the 3 attributes associated to the security status and get their value for my item, writing the attribute ids in my code too. But I dont like it I wanted to know if I missed anything and there was an obvious better way to do it ^^

I'll study your wall of text in detail tomorrow, but on top of my head: Dogma information from SDE and ESI is incomplete and difficult to understand. A long time ago, I tried to build a fitting engine from dogma data and I failed at understanding what stuff means&is supposed to do. SeAT's SDE is based on steve's exports. They are not 100% directly converted: https://www.fuzzwork.co.uk/2021/07/17/understanding-the-eve-online-sde-1/

yeah I read this part too and it stuck at the same place

directly, or correctly >:)

Ngl, the pyfa engine is not a bad idea

Those folks pulled off dark magic

And depending on your goal, using it, or atleast using it as an example is never a bad idea

https://discord.com/channels/821361165791133716/821361546608508938/1327570105189531710 @user_131158012831203328 I thought I was on to why we see so many failures by tracing through things using laravel telescope. My hunch was wrong though

yeah?

maybe I should make an effort :v and install laravel telescope on my install, then we could maybe sit down and look at things

might be easier to pin it down on a live install

Im running it on my prod server, but the problem wasnt what I though it was regarding middleware. I now think the problem just boils down to throughput

We cannot process enough of these jobs to get through them in a reasonable time because they exhaust too much of the rate limit.

As a simple method of printf debugging today I made the following patch: https://github.com/crypta-tech/eveapi/commit/8fd5d844f58a753ba702ad506de0c0d2b4797c9c

(though I backed out the Silenced change)

then I look at a Citadel job that failed after an hour

Grepping the logs for its jobID gives me the following

So 12 times in an hour that job was attempted and the middleware rejected it

On the 13th attempt, laravel failed the job

Ironically this patch only matters if it had of got past the ESI throttler

ah

so this job

it's looking up structure names

Yep

how does seat order structure lookups now?

I am not a good programmer, but I'm not completely lost when it comes to workflow diagrams 😄

because I'm seeing like a solid 390k failed jobs last 7 days and it's stable

and there's no reason for that

a unknown structure is discovered -> the structure is not in the citadel access block cache, and no other job for this structure+character combination has been scheduled in the last hour or so -> a job is scheduled -> the job is in the queue until the error limit checking middleware allows it to run, or laravel fails the job -> the job checks if the name hasn't been fetched by another character in the meantime. It also checks the structure access cache a second time -> do the esi calls -> if failed, update the stucture access cache, or update the database with the data

Hmmmm

No reason for the 390k jobs? or no reason for it to be stable. I expect that it would plateau at a certain number and be fairly stable. The number it stabilises at is a function of structures needing lookup, number of characters and schedule (which impacts the accesscache)

Its also very easy for the number of structures to balloon, all it takes is a character to have an asset there for example

We don't have the same number of new structures needing lookup every day, but maybe I misunderstood the logic

The problem is you cant get the information

The way ESI works, if the char doing the lookup doesnt have docking rights, ESI will return a 403

I'm aware

So we dont know what the structure is, we slowly keep trying to get it from all characters who have a link to the structure, and retry the characters that have failed on a slower cadence (set by the access cache)

If the character does not have docking access, aka 403, drop it on a no-retry list for like a month

We currently do for IIRC a week