Defining a locale using expressions? about datafaker HOT 9 CLOSED

no, it's not thread safe
however in datafaker initialization of Faker is very cheap in compare with javafaker it is about 100 times faster (we made some measurements for previous versions https://www.datafaker.net/documentation/performance170/#initialization)

so it's possible to create a faker per thread or even per task and it will be cheap (based on same measurements) performance of creation a new instance of faker is comparable with performance of simple expression evaluation. In case of complex expression initialization will be faster

from datafaker.

Do I need to construct a new Faker instance per call with the defined locale?

no, you can do it like that

Faker faker = new Faker();
faker.doWith(() -> faker.expression("#{Name.firstName}"), Locale.GERMAN);

from datafaker.

Please be more specific.

Each doWith call seems to mutate the current context, so it is not safe to use with multiple threads. I made this example:

fun main() {
    val faker = Faker()

    val threads = mutableListOf<Thread>()
    repeat(1000) {
        val locale = if (faker.bool().bool()) Locale.ENGLISH else Locale.JAPAN

        threads.add(Thread {
            val result = faker.doWith({faker.expression("#{Name.firstName}")}, locale)
            println("[$locale] $result")
        })
    }

    threads.forEach { it.start() }
}

There are many entries for Japanese locale containing english names and vice-versa.

no, it's not thread safe however in datafaker initialization of Faker is very cheap in compare with javafaker it is about 100 times faster (we made some measurements for previous versions https://www.datafaker.net/documentation/performance170/#initialization)

so it's possible to create a faker per thread or even per task and it will be cheap (based on same measurements) performance of creation a new instance of faker is comparable with performance of simple expression evaluation. In case of complex expression initialization will be faster

Will each Faker lookup using the resolver load the values from files into memory? In other words, should I be worried about memory consumption with instancing a new Faker per call?

from datafaker.

Will each Faker lookup using the resolver load the values from files into memory?

only if you request it.
It is lazy(after #33 and some other improvements), meaning that it loads files only for the provider which was invoked. If some providers are not invoked, then files for them will not be loaded. This one of the diff between datafaker and javafaker

from datafaker.

depending on your customer you might be interested in datafaker-gen(datafaker based configurable generator ) https://github.com/datafaker-net/datafaker-gen where it's possible to specify locale

from datafaker.

Amazing work with this library! Thank you for the doWith tip and the link, I'll check it out

from datafaker.

@snuyanzin the only thing I noticed is that this operation is not thread-safe. Is there a multi-threaded environment alternative?

from datafaker.

Please be more specific.

from datafaker.

Awesome, thanks for all the help!

from datafaker.