09:24 <FromGitter> <ndbroadbent> hello! I was just wondering what's going on with the Crystal VS Code extension? It's completely broken for me when I use "format on save", and this PR hasn't received any attention: https://github.com/crystal-lang-tools/vscode-crystal-lang/pull/99

09:26 <FromGitter> <ndbroadbent> I've love using Rubocop in my Ruby projects, even though it's incredibly slow (even with the rubocop-daemon wrapper.) So I was really excited to try out Crystal and get really fast auto-formatting (since it's compiled.)

09:27 <FromGitter> <ndbroadbent> do most Crystal developers not use the format option in VS code?

09:34 <FromGitter> <ndbroadbent> ahhh I see: https://github.com/crystal-lang-tools/vscode-crystal-lang/issues/84

09:59 <FromGitter> <ndbroadbent> would it be possible for all of the crystal-lang-tools repos to be officially supported and maintained by a team? It seems like vscode-crystal-lang is currently a project by a solo developer who doesn't have any time to work on it anymore

10:40 <kevinsjoberg> How do I pass command line arguments to my program when running "crystal spec <file>"? I tried doing "crystal spec <file> -- MYARG=1" but that does not seem to do it.

10:47 <repo> kevinsjoberg: how do you use them in your code?

10:47 <repo> oh to crystal spec...

10:47 <repo> not sure that works

10:48 <repo> you'd probably have more luck using env vars

11:10 <FromGitter> <ndbroadbent> I hope this will help! https://twitter.com/getdocspring/status/1201456721832505345

11:16 <FromGitter> <husam212> Hi, I'm trying a simple app using `kemal`, I want to render a template dynamically using one of the URL params, like the following: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=5de4f29ad64a052fb69f7bd8]

11:39 <FromGitter> <priit> Hi, I'm figuring out same goal, to render template dynamically. I looked into slang at the moment at https://github.com/jeromegn/slang

11:54 <kevinsjoberg> repo: Sure, I could try env vars but are you saying I can't pass a command line flag to my program when running it via "crystal spec"?

11:55 <kevinsjoberg> Do I have to build the binary for it to work then?

12:02 <FromGitter> <priit> @husam212: more reading at https://github.com/crystal-lang/crystal/issues/1942 @husam212

12:26 <repo> kevinsjoberg: well, spec has it's own arg parsing

12:26 <repo> so that will probably clash with yours

12:35 <FromGitter> <a8vbtb7p> Hello. I want to run a own Invidio instance. I must route it through socks5 proxy, how is it possible? Invidio otherwise wont work.

12:40 <FromGitter> <j8r> Invidio? Invidious you mean?

12:44 <repo> invidio.usa

12:44 <repo> -a

13:02 <FromGitter> <ndbroadbent> Hi everyone! Are there any other Crystal sponsors in here? Please let me know if you might be interested in pooling our resources and sponsoring incremental compilation for the Crystal compiler: https://forum.crystal-lang.org/t/sponsorship-for-incremental-compilation/1428

13:08 <FromGitter> <j8r> Unfortunately incremental compilation is not for the near future, because it requires quite big breaking changes in the language design

13:08 <FromGitter> <j8r> It's not only technical

13:10 <FromGitter> <j8r> for now the best bet is to split a huge application in smaller ones, that's why Crystal is perfectly fitted for micro-services - despite having static linking broken :(

14:13 <FromGitter> <asterite> I was going to reply something like that, but I didn't because we think incremental compilation might not be possible without language changes, but it's just a thought. We need to investigate it to draw good conclusions. So I think having an opportunity to do that would be fantastic.

14:13 <FromGitter> <asterite> That is, let's not close the doors before they are open

14:17 <FromGitter> <ndbroadbent> I would love to hear more about the reasons why incremental compilation will be difficult! I've been spending some time thinking about it, and I can't really see why it's too difficult

14:17 <FromGitter> <ndbroadbent> even if it's not perfect, I'm sure there are tons of things that can be cached in a simple way

14:18 <FromGitter> <ndbroadbent> just based on file mtimes and AST nodes

14:22 <FromGitter> <ndbroadbent> whoops I'm just reading this post from almost 4 years ago: https://crystal-lang.org/2015/12/24/the-future-of-crystal.html

14:22 <FromGitter> <ndbroadbent> haha sorry, I'm sure a lot of thought has gone into this

14:26 <FromGitter> <ndbroadbent> are there any more recent posts and articles I should read?

14:35 <FromGitter> <j8r> @ndbroadbent I would like to know, any good reason not to go to API <=> Front design?

14:35 <FromGitter> <j8r> because you will rewrite your application

14:36 <FromGitter> <ndbroadbent> oh yeah, that could be an option, if I move all my front-end into a React SPA with TypeScript

14:37 <FromGitter> <ndbroadbent> but the static type-checking for Lucky views also sounds really nice: https://luckyframework.org/guides/frontend/rendering-html

14:37 <FromGitter> <j8r> I you want perf, and less compilation time overall, i think it's a good option

14:38 <FromGitter> <ndbroadbent> but I do still have a lot of models and backend code for my API, and background job workers, etc.

14:38 <FromGitter> <ndbroadbent> i've been talking to the creator of the hexapdf library, who might be interested in working on a Crystal port: https://github.com/gettalong/hexapdf

14:38 <FromGitter> <j8r> Ok, but you have to rewrite everything :/

14:42 <FromGitter> <ndbroadbent> so one of the main reasons I'm super interested in Crystal is because I'm working on some PDF generation software, which is originally written in Ruby. I've been using the Prawn PDF library, which is basically unmaintained and full of bugs and unhandled edge cases. I've been working with some contractors on some of these PDF bugs and crashes, and I realized that I've probably spend over $10k just fixing

14:42 <FromGitter> ... mostly type errors that would have been instantly caught by Crystal

14:42 <FromGitter> <ndbroadbent> so porting my PDF generation engine to Crystal (and running the job with sidekiq.cr) will be a huge win

14:43 <FromGitter> <ndbroadbent> so I think that will be a great start, and can be done independently

14:44 <FromGitter> <ndbroadbent> then later on it would be great to also rewrite my API endpoints in Crystal. So there are some ways that I can keep things separate

14:45 <FromGitter> <Blacksmoke16> JSON API?

14:45 <FromGitter> <ndbroadbent> i'm using OpenAPI / Swagger

14:46 <FromGitter> <ndbroadbent> also very interested in porting the rswag gem to Crystal, because that's been really nice to work with (https://github.com/rswag/rswag)

14:47 <repo> ndbroadbent you can try out my shard :) https://github.com/repomaa/open_api.cr

14:48 <FromGitter> <ndbroadbent> awesome!

14:48 <hightower4> ndbroadbent interesting, do you think openapi will continue to prosper when there is now graphql?

14:48 <repo> i think so

14:50 <FromGitter> <ndbroadbent> I think they can both be useful for different cases. graphql is great for data-centric APIs where people need to read and query a lot of data, but I really prefer REST / OpenAPI when I'm building an API-as-a-service

14:50 <repo> currently our architecture is multiple micro services with rest api providing index actions that allow filtering with ids (amongst other things), and then we have a thin graphql layer on top of that which uses the pretty awesome dataloader js lib to optimize the requests to the api

14:51 <repo> the tooling for graphql is pretty great on js altough i really don't like writing it

14:51 <repo> namely apollo server

14:52 <repo> but dataloader is really a great solution for connecting graphql to rest apis

14:53 <repo> something that would translate into hundreds of requests suddenly translate into 12 or so

14:53 <repo> and it keeps the rest api's very straight forward

14:54 <repo> no magic needed

14:54 <repo> no relation fetching

14:55 <repo> but it's crucial to provide index method which take an array of ids as a filter query

14:55 <repo> *an

14:56 <repo> otherwise dataloader can't utilize caching

15:00 <FromGitter> <Blacksmoke16> normally id say "check out Athena" if you're making a JSON API, however it's going thru some heavy refactoring ATM; so probably not the best idea...unless you want to be *bleeding* edge ha

15:44 <FromGitter> <yxhuvud> I rails I found the json api implementations to be too opinionated to work well with our use case so we rolled our own.

15:46 <FromGitter> <j8r> @ndbroadbent you can also find a C library, and generate bindings

15:47 <FromGitter> <yxhuvud> but then we use the standard more to shortcut format and data layout discussions rather than as way to automate integration.

16:18 <FromGitter> <kinxer> @Blacksmoke16 Surely there are older releases of Athena that would still work well?

16:18 <FromGitter> <Blacksmoke16> the current one would yea

16:19 <FromGitter> <kinxer> Is your major refactor going to break a lot of the API?

16:19 <FromGitter> <Blacksmoke16> yes

16:19 <FromGitter> <Blacksmoke16> like almost all of it :p

16:19 <FromGitter> <kinxer> Ah... Gotcha. So probably it's not a good idea to use an older release.

16:20 <FromGitter> <Blacksmoke16> yea, switching controllers from classes to structs, and breaking DI changes, and changing how i handle events (moving away from `HTTP::Handler` in favor of https://github.com/athena-framework/event-dispatcher style

16:23 <FromGitter> <Blacksmoke16> granted a lot of the changes are internal, mostly minor breaking changes for the end user. But still prob not worth setting something up just to have to redo some of it again later

16:27 <FromGitter> <Blacksmoke16> id be happy to go over the current plans/status if you/anyone else has any feedback. Is hard designing everything on your own ha

16:34 <FromGitter> <tenebrousedge> Kai recommends breaking all the things without prior notice

16:34 <FromGitter> <tenebrousedge> that's how you get the most feedback

16:34 <FromGitter> <Blacksmoke16> hehe, granted i wonder how many people actually read the changelog before updating

16:34 <FromGitter> <Blacksmoke16> *guess we'll find out*

16:35 <FromGitter> <kinxer> I mean, you'll increment the minor revision number, right? That should at least prevent anyone with ` ~0.X.*` versions from getting the newest changes accidentally.

16:35 <FromGitter> <Blacksmoke16> yea ofc

16:35 <FromGitter> <Blacksmoke16> `0.7.0` to `0.8.0`

16:42 <repo> i'll put Athena on my list

16:43 <repo> last time i checked nothing was quite fitting my requirements so i went and wrote mostly from scratch using crecto, router.cr and paramnoia

16:44 <FromGitter> <Blacksmoke16> what were your requirements?

16:44 <repo> i needed crecto and i wanted type safe params

16:45 <repo> so far crecto is the most advanced orm imo

16:45 <FromGitter> <Blacksmoke16> is independent of any particular ORM, however some features would require the ORM to implement some specific methods if it doesnt have them already

16:45 <repo> i had a lot of really complex joins and queries that none of the big ones managed

16:45 <FromGitter> <Blacksmoke16> i.e. like `Model.find id`

16:46 <repo> (without resulting to raw sql)

16:46 <FromGitter> <Blacksmoke16> is it back? last i heard it was abandoned

16:46 <repo> yeah it's back

16:46 <FromGitter> <Blacksmoke16> ah nice one

16:46 <repo> but that was tough when fridgerator announced it's abandoned :D

16:46 <repo> i had just built a multi-thousand-line codebase on it :D

16:47 <FromGitter> <Blacksmoke16> xD that would do it

16:47 <repo> but now i'm happily running it on 0.31.1

16:47 <repo> i really really like avram

16:48 <repo> but it doesn't even support OR

16:50 <FromGitter> <Blacksmoke16> i mainly been going with Granite, fits into the annotation approach similar to what Athena does, afaik its the only one that allows you do apply annotations to the underlying ivars

16:51 <repo> afair granite wan't typesafe

16:51 <FromGitter> <Blacksmoke16> typesafe queries?

16:51 <repo> yeah

16:51 <FromGitter> <Blacksmoke16> yea dont think that part is

16:52 <repo> oh CrSerializer looks really nice!

16:52 <repo> good job!

16:54 <FromGitter> <Blacksmoke16> thanks :)

16:56 <FromGitter> <Blacksmoke16> serialization side of it is mostly done, still some things to iron would with the other direction

16:56 <repo> cool

16:56 <FromGitter> <Blacksmoke16> er no i think that was finished too

16:58 <FromGitter> <Blacksmoke16> really just need to start migrating these shards to the org now

18:27 <Xeago> Hiya, I'm looking to spend some hobby time with crystal this holiday period. I saw references to use vscode and native-debug. I've installed `faustinoaq.crystal-lang` into vscode but get nowhere near the developer experience shown in the readme for the extension.

18:28 <Xeago> I created appropriate task and launch config files, which do build but then don't actually launch a debugger.

18:31 <Xeago> I can run code using the play button in vscode, but debugging is futile. I've checked, and llvm is installed.

18:35 <FromGitter> <tenebrousedge> you're working off of this guide (https://github.com/amberframework/docs/blob/master/examples/crystal-debug.md)?