Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
Tonni Tølbøll Lund Aagesen
@ta
yes, was about to write that solution :)
can there be an argument for having nested structs inherit from parent struct?
Nikita Shilnikov
@flash-gordon
we could have a means for that
I'm not sure what's the best here but having something would be nice
I came across this pattern a copule of times
Tonni Tølbøll Lund Aagesen
@ta
Thanks for helping out
oh btw, can a nested struct be optional like a nested schema?
Nikita Shilnikov
@flash-gordon
you mean absence of key or nil as a value?
Tonni Tølbøll Lund Aagesen
@ta
I have Types::Hash.schema( ... ).optional.default(nil) in some structs
Nikita Shilnikov
@flash-gordon
class MyStruct < Base
  attribute :id, Types::Strict::Integer
  attribute :data, Class.new(Base) {
    attribute :code, Types::Strict::Integer.optional.default(nil)
    attribute :msg,  Types::String.optional.default(nil)
  }.optional
end
that's the closest thing I can think of
and that's not good enough
feel free to post both suggestions to GH and we'll sort it out eventually
Tonni Tølbøll Lund Aagesen
@ta
will do, thx
Tonni Tølbøll Lund Aagesen
@ta
oh my, what one can learn from reading the docs more closely: Types::Hash.schema(name: Types::Strict::String).with_key_transform(&:to_sym) :)
Nikita Shilnikov
@flash-gordon
yeah, if you don't need structs you can use schemas, they also support kind of inheritance
I prefer DSL for describing structs, this is basically it
Jaromír Červenka
@Cervajz
Off topic: Anyone heading to ElixirConf in Prague? :)
Oskar Szrajer
@gotar
Hi, I'm trying to validate something like this in dry-v. a field values is any array of strings, if inside those array there is a value 'foo' it must validate presence of 'bar', 'baz' too, so something like
values.include?('foo') > values.include?('bar') & values.include?('baz')
is it possible? I don't have any good idea how to achieve this
Vasily Kolesnikov
@v-kolesnikov
I would prefer to avoid that validation at all, even if dry-v could do it. It sounds strange to have data like unstructured array of strings. Could you give a real example of usage the data like that?
Oskar Szrajer
@gotar
User select answers in a form, it's a checkbox (multiselect) is he select one from them he must select some others.
I found some solution, just a custom predicate, works ok-ish ;]
Vasily Kolesnikov
@v-kolesnikov
:ok_hand:
Oskar Szrajer
@gotar
and I'm building this for existing form, cannot change logic/form there
Ariel Valentin
@arielvalentin
Hello friends. Pitor recently invited me to collaborate on dry-system-rails. I have kept most of my comments in the discourse forum or pull requests. Are those still the best place to discuss design/code changes? What is appropriate to discuss here?
Tim Riley
@timriley
@arielvalentin both of those places are better than here, yep. I’d keep doing what you’re doing.
Ariel Valentin
@arielvalentin
Thank you @timriley !
Trey Pendragon
@tpendragon
Hey all. I was wondering if there's a plan for a 1.0 release of the dry ecosystem?
Our community has started using it, and we've had a system in production using it instead of ActiveRecord and the like for the last year, but it'd be nice to be able to rely on semver.
Nikita Shilnikov
@flash-gordon
@tpendragon is it that important? :) A few things to keep in mind
the API is pretty stable already and we try to make all the changes in a backwards compatible manner
Trey Pendragon
@tpendragon
I agree, you've done a great job. Is it a problem to just cut what exists as 1.0.0 and keep on doing that?
Nikita Shilnikov
@flash-gordon
second, we coordinate our releases with hanami's ones, that means if for any reason we'll need breaking changes (unlikely) this will give us some freedom. Still, we follow semver so this shouldn't be a problem
but we do have checklists for 1.0s
Trey Pendragon
@tpendragon
Working for Universities, we tend to put a lot of stock in specifications, so I just know it would make a number of my colleagues more comfortable were things like dry-types and dry-struct to have the commitment behind them that comes with 1.0. Otherwise, while you've done a great job, technically you're outside the realm of responsibility yeah?
If you're not comfortable with it I understand, I know it's a daunting thing. Do you have a link to the checklist of requirements somewhere? Maybe I can help.
Nikita Shilnikov
@flash-gordon
as I said, it's a matter of coordination, otherwise I'd push 1.0 and call it a day :) I don't think there's a big issue with how you number your gems as long as you follow semver. What I can say, it's dry-types is expected to be the next 1.0 release since it's mature and stable. But we do have plans to introduce some (non-breaking I hope) changes before that.
Trey Pendragon
@tpendragon
Okay. :) I'll keep an eye on it. We're about to release some code that depends on 0.13 of dry-types, so I look forward to seeing what's coming up.
Thanks!
Manic Chuang
@manic

Hi guys, it seems that I maybe encounter a bug of dry-types SUM, but I'm not sure. Could you help me on this?

It happens when I try to verify a object which should be a hash or a string

test = Dry::Types['hash'].map(Dry::Types['symbol'], Dry::Types['strict.string']) | Dry::Types['strict.string']
test[a: 'b'] # passed as expected
test['a' => 'b'] # should fail but pass!
test['test'] # passed as expected
test[12345] # should fail but pass!

I'm wondering if there is anything that I didn't notice or I use the wrong way to build my rule. Thanks :)

Tim Riley
@timriley
@manic I tried this, using “strict.symbol” instead of “symbol” (which IMO you probably want) and it still fails in the way you’re seeing. Could you please file a bug?
Manic Chuang
@manic
@Tim Sure, thank you
Manic Chuang
@manic
@timriley dry-rb/dry-types#280 here it is.
Pablo Crivella
@pablocrivella
Any idea how could i make this Map type also accept nil as values?
Links = Map(Symbol, String | JSONAPI::Types::Link)
This works but i’m not sure if its the correct way:
Links = Map(Symbol, String | JSONAPI::Types::Link.optional)
Nikita Shilnikov
@flash-gordon
no idea, probably related to the issue above
Jeff Dickey
@jdickey
Hi, folks. Question about dry-configurable. Do I understand correctly that "pre-processing" with a block is done once and memoised? Does defining a block automatically imply reader: true ?
Tim Riley
@timriley
@jdickey providing a processor doesn’t have any effect on whether the setting comes with a reader or not