Test helpers in preloaded snippet of C translations

[hobovsky]

I have a question I have been bothered with for some time, and I was asked to create a public discussion on the topic, so here it is:

When new C translations appear, they sometimes happen to have some helper function for tests in Preloaded part. It's especially common for some particular C and C++ translators, who put there some boilerplate (helper functions for tests) shared between sample tests and submission tests. I am not a big fan of this approach, for two reasons:

• Preloaded snippet is usually meant to contain code shared between user solution and tests, and not between tests and tests.
• Test helpers are not visible to users, and as a consequence sample tests cannot be just copied to local environment and run, making training in local IDEs a pain.

I solve almost all non-white kata in local IDE and I don't really use Codewars trainer a lot for anything else than whites. Being unable to train on C kata locally in an easy way triggers me a little every time.
On the other hand, I can totally see why helpers are in preloaded code: having them separately in sample tests and submission tests causes code duplication, and, as a result, maintenance issues arise.

There are several possibilities to mitigate problems caused by test helpers placed in preloaded code:

• Allow them, because problems stated above are not problems at all, and no one trains on C kata locally.
• Allow them and ask for the problem of inaccessible preloaded snippet to be fixed by system (for example, add a view to kata trainer which shows preloaded snippet) - as far as I know this solution is not easy due to some technical obstacles.
• Allow them and ask authors to put preloaded code in description.
• Discourage/disallow them and ask authors to put test helpers in corresponding snippets (sample tests and submission tests).

Nowadays the direct cause of my doubts are C translations from user @trashy_incel who publishes really many C translations, and which are of reasonable quality. My problem with them is that they (ab)use preloaded a lot, basically rendering kata impossible to train on locally.
I would like to approve most of them, but the preloaded thing keeps bothering me. But we don't have any good alternative for now :/

[hobovsky]

I already asked the question in a smaller group some time ago, and some answers were:

kazk:

I prefer not having test helpers in preloaded, but I don't have a strong opinion for C/C++.

monadius:

I rarely use kata tests directly for local development (never for NASM/C/C++). For me it is fine to have some boilerplate code in Preloaded. Although, I prefer to keep Sample Tests as simple as possible. Very often, Preloaded is not required for simple tests.

Unnamed:

Duplicating the bolierplate to both test snippets?

Maybe this... But if a translation author refused to change minor things like this, I'd probably just approve it.

kazk:

Personally, I agree with monadius and Unnamed. What are they doing to require that much boilerplate code in sample tests? I don't like when a kata uses custom wrapper for no good reason. Keep sample tests simple and easy to understand, so solvers can modify them. I also think duplicating some code is fine.