Re^9: use feature 'postderef'; # Postfix Dereference Syntax is coming in 5.20 (hate)

[tye]

And I don't think I should have to monitor the p5/porters lists or lose any right to voice feedback.

If you perceived that I in some way was trying to rescind or even object to your right to voice feedback, then that is unfortunate (and I'm at a loss as to what I said to encourage such an interpretation).

If you thought the prior proposals for this were found by me on p5p, then you should know that I rarely read much of anything from p5p and the prior mentions (IIRC) that I saw were all found right here on PerlMonks.

That you refer to reasonable criticisms as "hate"

I wasn't questioning the reasonableness of any particular criticism. I was expressing surprise at the relative avalanche of nearly exclusively negative responses (and, to a lesser extent, the relative vehemence of some of it). Hence my mention of my prior (and highly contrasting) experiences with responses to nearly identical proposals.

Indeed, the level of unanimity to the "hate" (using the relatively recent definition) that I perceived made me reluctant to contribute to the thread. And that made me hope that somebody would break the pattern such that perhaps others would feel comfortable expressing differing opinions. But then I eventually realized that I had things to contribute beyond just my personal opinion. But I suspect my contribution came too late (after the thread was already crowded) and so won't serve well to encourage like-minded responses.

this introduces feature duplication

Yes, having more than one way to do it is surely counter to Perl's whole philosophy. q-:

that isn't backwards compatible

I have yet to see the backwards compatibility problem(s). The closest I've seen is what I mentioned, that the parsing of $obj->$* changed. But the prior meaning seems clear to have zero chance of having been used.

I also have to assume it comes with a performance hit, however small

I see no reason to make such an assumption. I see every reason to expect that, if there is some theoretical performance hit, then it is astonishingly trivial.

and an ongoing maintenance cost.

It is experimental so there might be zero long-term maintenance cost. The experiment might have been required just for the sake of eliciting all of the "hate" that wasn't forthcoming (that I saw) despite having floated the idea for many years. :) But at least this is an argument I can see some merit in. Though I suspect that a project that leans much on such a concern will soon be a relatively dead project.

"Is it worth it?" is the real question.

I suspect that the more consequential impact will be upon the attitudes of contributors or potential contributors to p5p.

"More experiments" seems the wiser route for the long-term viability / vitality of Perl 5, IMHO.

I'm glad to see more progress in making postfix dereferencing complete. The syntax for the "slice" versions has been stable for many years. The syntax for non-slice versions appears to still be undergoing experimentation (based on recent commits I noticed). So at least this new increase in the impact of the experiments has had the benefit of greatly increasing the visibility of the concept and the quantity of feedback. Even if the experiment is eventually completely reverted, the resulting increased discussion will likely have been beneficial.