Meta::Helper: be more calling-context aware.

[mwmiller]

Since we're offering html_enc as an alternative to encode_entities
it should behave in the same way when supplied a scalar and expecting
scalar output. Mostly we're concerned about when they want scalar
output, so we'll just encode the first element off the supplied array if
that's how they've called us.

uri_esc is similarly updated to provide similar semantics.

cc: @mintsoft

[mwmiller]

An alternative is to concatenate the string in the scalar case. I'm not too bothered going either way.

[mintsoft]

@mwmiller I think I'd expect it to concatenate if going from array to scalar context for situations like:

return "and your answers are: ".html_enc(@answers)

Edit: on reflection, isn't that useless anyway?! It'd be far better to just do html_enc(join " ",@answers)

[mwmiller]

@mintsoft I can make arguments both ways. I'm honestly not sure which way provides "less astonishment." I'm mostly worried about the conversion of the single element to scalar case, so I don't have a well-formed opinion.

Edit: Also, I went ahead and changed it based on your preedit comment, because I had no strong opinion.

[mintsoft]

@mwmiller on reflection this morning I think my original thoughts make the most sense. I'd be weird to pass an array into something and have it appear to randomly lose data by only using the first element of the array.

So I think it's now correct

[mwmiller]

have it appear to randomly lose data

This was what made me decide to switch it based on your original comment. I figure at least it's obvious what happened with all of the data if you call it "incorrectly."

[moollaza]

This is awesome guys, nicely done. 👍
Regarding your discussion, I think the current implementation is best and makes it most useful for the context we tend to use this function in. I guess this all might change though when we have the Goodie templates?

[mintsoft]

@moollaza very possibly; but I think it probably depends on how they're implemented

[mwmiller]

@moollaza I don't know how it might change with Goodie templates, but I'm pretty deeply involved there, so I'll keep myself in the loop.:grin:

[moollaza]

didn't know wantarray existed. Very cool

[kablamo]

I believe wantarray is often (perhaps always?) astonishing in some context. This is the YAPC talk that convinced me. Its kind of a fun talk actually:

http://aaroncrane.co.uk/talks/calamitous_context/
https://www.youtube.com/watch?v=NbP0AtEaxBU

Its really lame of me to make a comment after this has been released and its not really a big deal anyway. I just thought I would throw this idea out to consider for the future.

[mwmiller]

Its really lame of me to make a comment after this has been released

@kablamo No, it's actually a very good point! I don't think it's particularly dangerous or horrible in the proximate case, but we should always be wary of context-sensitive code. I think the reason I am more or less ok with it here is because it directly on the provided arguments, not some hidden state which would make it more difficult to reason about what's happening. I might well be wrong about how this is better. 😁

Thanks for the link to the talk. I found it quite interesting.

[mintsoft]

@mwmiller you mentioned the goodie templates before, is there a place where that work is being done? Is there anything I can do to assist?

[mwmiller]

@mintsoft There's nothing much to see right at the moment. I have a working prototype, but everything interesting happens on the closed-source side of the house at the moment. I expect in the next week or two the changes will appear to leap fully-formed into a repo near you. At the point, I look forward to your help in identifying edges I have missed as we go about converting goodies. 😁

[mintsoft]

@mwmiller cool, just lemme know if I can be of any use 👍