methods implemented in terms of _get_or_search constraint author, module, distribution, and release names too much

[rafl]

_get_or_search, which is used in the implementation of several other methods, will only delegate to ->_get for arguments that are both defined and match the regular expression /^[A-Za-z-:]+$/.

There are valid module, distribution, and release names that aren't matched by the above regex. There's also invalid module names that are matched by it.

It would seem appropriate to me to always dispatch to ->_get if the argument is both not a reference and defined at the same time.

[xsawyerx]

The only downside of validating this in the module is allowing incorrect requests to be sent to MetaCPAN and have them fail there. If we can prevent the overhead, why not?[1]

Imagine a user who wrote a script that sends a lot of requests and it had a bug in it which sends incorrect distribution names. They will spam MetaCPAN without any value in it anyway.[1]

[1] unless, of course, I don't understand how you want to approach this.
[2] not that we would encourage spamming MetaCPAN, and I'm aware there are limitations on responses and perhaps throttling even.

[rafl]

I don't really have a strong opinion on MetaCPAN-Client validating things locally or not. I'd merely like it's validation, if there is any, to permit all valid values. Right now it's too restrictive.

[xsawyerx]

I understand the restrictiveness. What I'm trying to follow this up with is: do you think it's better to make the validation more accurate, or to remove it entirely?

[rafl]

I don't have a strong preference between those two options. If I was to implement a fix for this issue, I'd probably do it by removing the existing validation. That seems a lot simpler than coming up with code that will accept all valid module/dist/etc names while still being restrictive enough to be useful as a validation.

Trying to help MetaCPAN to not be flooded with broken requests because of programmer errors is certainly a nice thing to do, but that's a kind of problem MetaCPAN will still have to deal with even if MetaCPAN-Client validates things locally, so I'm not sure effort put into validations is gonna be terribly well spent.

[xsawyerx]

I'm convinced. Fixed, merged, pushed.

Will be in next release.

Thanks. :)