printf "%d\n", $x;  # interpolate %d?
printf '%d\n', $x;  # don't replace \n?
printf "\%d\n", $x;  # ick
printf '%d'."\n", $x;  # double ick
[download]

If % interpolated, printf would be kind of hard to use, wouldn't it?

This does not explain why % does not interpolate; rather it shows that the choice of % as a meta character in printf is a bad one given that % is a sigil. If pritnf used a different meta character (^ for example), then this would work:

printf "^d\n", $x;  # no problem
printf "%d\n", $x;  # interpolate, no problem
printf '^d\n', $x;  # ok
printf '%d\n', $x;  # does not interpolate, ok too
[download]

So, you can make % interpolate without making printf any harder to use (using ^ instead of % is not harder, just different), you just rather keep printf the way it is (so that people don't get angry).

Well, you should realize that printf was in Perl before hashes were first class objects with their own sigil.

Abigail

This does not explain why % does not interpolate...

I think it shows why making hashes interpolate would be a bad idea (given the current state of affairs), which amounts to the same thing.

... rather it shows that the choice of % as a meta character in printf is a bad one given that % is a sigil

Unfortunately, it's hard to make all those design choices up front, and now we're pretty much stuck with them. Larry could decide to change the printf escape character for Perl6. It would make it difficult to mechanically translate Perl5 programs that use non-constant printf format strings to Perl6, but maybe he'd find that acceptable. Changing the hash sigil at this point would be a much bigger problem.

printf '%d\n', $x;

To make this work, you'd need to have printf itself interpret the \n escape. That could be done, but it makes me nervous. Newbie programmers seem to have a hard enough time understanding escape sequences without muddying the water this way. I can just hear it: "Why does printf 'foo\n' work, but not print 'foo\n'?" Plus, if you actually wanted to printf a literal backslash, you'd need to put four backslashes in your code. Yuck.

Maybe there could be a new printf directive for a newline.

printf '%d%n', $x;

Maybe it's tricky to find the right representation. Do you want just the keys? Just the values? Keys mixed with values? Does order matter?

Converting an array to a list is pretty sensical, but converting a hash to a list is a little different. (Think in terms of pairs.)

Oh dunno - we already have a definition for converting a hash to a list and an interpolated list array to a scalar. So I just wonder why a hash shouldn't interpolate as a list array. Perl5 doesn't have the same ideas about pairs that Perl6 does so at this point I'm not concerned about what to do with pairs because they don't exist.

Lists don't interpolate. (How could they? They don't have names, let alone a sigil).

There's limited value in having hashes interpolate directly; it's not as useful as interpolation of scalars and arrays. It would be very annoying though, as you would not be able to write

    printf "%d %s %f\n", $int, $str, $float;
[download]

because that would try to interpolate three hashes.

However, it's still relative simple to interpolate a hash:

    print "@{[%hash]}";
[download]

Abigail

Historicly, BTW, note that the interpolation of scalars has been there since the beginning (I think), whereas interpolation of arrays is rather new (5.0 partialy, normalized somwhat in 5.6, thanks, toma).

I'm almost certain that interpolation of arrays was available in perl4. Hmm... it was. I just tested it with perl 4.036.

See the first paragraph in this document describing traps when going from perl4 to perl5.

@ now always interpolates an array in double-quotish strings.

Note the word "always". It used to be so that perl4 only interpolated arrays where they existed before being used in a string.

Also, adding 1..10 to an array vs. 11..20 will still return them in a predictable (a < b) order. Doing that for a hash will not return a predictable ordering.

This matters for testing purposes - I now cannot depend on interpolation to always give me repeatable answers for testing. But, the only thing I can't depend on is hashes. Try explaining that to a newbie.

------
We are the carpenters and bricklayers of the Information Age.

Don't go borrowing trouble. For programmers, this means Worry only about what you need to implement.

Please remember that I'm crufty and crochety. All opinions are purely mine and all code is untested, unless otherwise specified.