[enghitalo]

This PR

fn test_strict_check() {
	assert strict.strict_check[StructTypeAndOptionType[string]]('{"val": "","val": ""}') == strict.StructCheckResult{
		duplicates: ['val']
		superfluous: []
	}

	assert strict.strict_check[StructTypeAndOptionType[string]]('{"val": "","val2": ""}') == strict.StructCheckResult{
		duplicates: []
		superfluous: ['val2']
	}
}

Next PRs

strict_decode

• Stop on first error to be performative

struct StructType {
mut:
	val        string
}

fn test_superfluous() {
	json.strict_decode[StructType]('{"val": "","val2": ""}') or {
		assert err.msg() == '[x.json2] useless data detected. Try pruning superfluous fields shown in err.superfluous_fields()'
		assert err.superfluous_fields() == ['val2']
		return
	}
	assert false
}

prune

struct StructType {
mut:
	val        string
}

fn test_superfluous() {
	json.prune[StructType]('{"val": "","val2": ""}') == '{"val": ""}'
}

Performance improvments

[enghitalo]

How can this CI fail can be related to this PR?

[ttytm]

One thing that comes to my mind would be to get the duplicates and superfluous in one loop. Both are recursively called, so it could be worth a try using one instead of multiple consecutive loops. Did you try that?

[enghitalo]

I will try it in next PR. Thank you!!

[ttytm]

How can this CI fail can be related to this PR?

Good work! It's likely not related 🙂 👍

[enghitalo]

Thank you for review @ttytm !

[spytheman]

I do not understand, what is the use case of this submodule?

[spytheman]

I think we have a bug in the compiler, that allowed non public functions to be called by the test, without it being marked as an internal one.

[enghitalo]

I do not understand, what is the use case of this submodule?

Many of the companies I worked for had slowdown issues due to excessive unnecessary data sent through the back-end of web applications. Using a library like this, on the front-end, for example, helps keep the back-end code fast, efficient, safe and with better-defined routes. Which results in a lower cost for projects in production

[spytheman]

I still do not understand the usecase 🤔 .

Lets take for example a web app, and let it have:
a) a JS client running in a browser,
b) a backend webserver implemented in V, connected to
c) a DB, say MySQL.

The communication between the JS client and the backend in V, happens by sending JSON messages.

Where would such a module be used?
aa) when the V webserver sends JSON to the JS client?
bb) when the V webserver receives JSON from the JS client?

If aa) is the case, could not the webserver just not generate the additional keys, in the first place, and that would have been caught in code review?

If bb) is the case, what would the webserver do, when it detects it? Log that it found duplicated keys? Who will read those logs? Or will it stop processing, until the JS client is fixed?

[spytheman]

I also could not understand how is that related to routing at all 🤔 .

[enghitalo]

Hi. Sorry the delay. I will answer everything As Soon As Possible

[enghitalo]

I also could not understand how is that related to routing at all 🤔 .

It is very common for some developers to use a single route for multiple purposes, what is no good. For example, a route that returns all the products in the database, sometimes even with relationships, when, in many cases, they do not need all this information. Use the strict library to check that there is no unnecessary data, to perform end-to-end tests, send a log, or even return an error. It would help to create and maintain a final product with greater quality and performance, create a quality metric for the development team, facilitate code review...

[enghitalo]

I still do not understand the usecase 🤔 ....

More the bb) case.

It will depend on how rigid the team is. Personally, I would use Log Management or apply it to tests. Since, for me, it wouldn't be very interesting to break the application with errors that could be random (But it also depends on how these errors will be handled)