Hi guys, I have some trouble using dry-validation. I am trying to ensure that some ids I receive from my API are actually valid records in my database. In my form I have this :
property sender_ids
required(:sender_ids) do
filled? & array?
end
rule(valid_senders: [:sender_ids]) do |sender_ids|
Contact.exists?(id: value(:sender_ids))
endBut it raises the following error :
POST /letters with valid data returns the new letter
Failure/Error:
rule(valid_senders: [:sender_ids]) do |sender_ids|
Contact.exists?(id: value(:sender_ids))
end
NoMethodError: undefined method `with' for false:FalseClass
# ./.gems/gems/dry-validation-0.10.4/lib/dry/validation/schema/value.rb:96:in `rule'
# ./app/forms/letter/create_form.rb:31:in `block in <class:CreateForm>’Do you have any idea ?
optional(:attribute_name) { eql?('') | none? | (str? & format?(/\w+/) & max_size?(10)) }By calling
MyScheme.call(attribute_name: 'some invalid value').messagesI'm getting errors as a single string
{:attribute_name=>["must be equal to or must be a string or is in invalid format or size cannot be greater than 10"]}I need to have one error message for all kind of errors, like "Attribute name must contain < 10 symbols from the following list: ..."
Is it possible?
Dry::Validation::MessageCompiler I guess it doesn't support any options for switching generating strategies...
| (Or) is the only node in produced AST that gets converted to a non-array value
&, > or ^ my errors messages are not joined. This code is a problem for me - https://github.com/dry-rb/dry-validation/blob/master/lib/dry/validation/message.rb#L35 - and I have no idea how to solve it without patching
def my_predicate
return eql? || none? || ... #this is pseudo code
end
my_predicate: 'custom message'
optional(:attr) { my_predicate }
hi! I'm running into a struggle with dry-transaction and I wonder if it is something any of you have run into.
1)The interface expected is always in the form call(input) but since we have been using dry-initializer a lot, our callable objects are initialized with the input, and called with call() without any arguments.
I can wrap the registered methods so they get called correctly but seems a bit of an overhead. Step adapters don't really solve anything since they adapt the handling of the output but not the way the method is called.
Is there any way to streamline this? Has any of you felt this need before?
#call-ing them once.
#call is receives the variable input.
#call on it many times over.
#call method that does that to_either conversion for you
newbie here. I watched @timriley's Singapore presentation and really liked the functional-ity and separation of concerns. I'm looking for some convention to surround me as I get started with these new-to-me dry and rom capabilities. Something like a "click here to download Trailblazer with skeleton support for a Riley-like dry/rom app".Is there any getting-started resource like that? Maybe @fran-worley can recommend where someone new to both dry/rom and Trailblazer should start if they have no legacy app to evolve?
I'm really interested in the ROM Relations concept that allows architect/developers to work with objects that look like business objects rather than detailed database scaffolding because the kind of applications that I work on have complex object-with-object relationships and all relational database repositories eventually so pollute the network-graph business concepts that the business services often have to be simplified or be reduced in flexibility if the relational-model details are implemented in the business objects themselves. Therefore, the dry/rom architecture of separating out associations as proper objects, of separating out repository query handlers, of separating out functional dependency-injected command objects, from the business objects themselves is intoxicating.
true? https://github.com/dry-rb/dry-logic/blob/bed90e00489e4c133499126428ce1dc0f7beb9f6/lib/dry/logic/predicates.rb#L173-L175