Toward the end of the Comma Operator section in perlop it says:

If the argument on the left is not a word, it is first interpreted as an expression, and then the string value of that is used.

which is consistent with observed behaviour.

In which case a word must be something other than a sequence of "word characters". Is word defined somewhere?

update: I'm guessing a word must begin with an alphabetic character or an underscore so that anything beginning with a digit is not a word.

update: Looking at perl source (toke.c) it appears that anything beginning with a digit is scanned as a number while anything starting with an underscore or alphabetic character other than 'v' or 'x' is scanned as... there are quite a few possibilities here, one of which is "a word before a => operator". There is some additional testing done for 'v' and 'x'. I don't follow it all, but suspect 'v' might be the start of a version string, handled somewhat differently, and 'x' followed by a digit might be a repeat operator, depending on context. Otherwise 'v' and 'x' are handled the same as all the other alphabetics.

In conclusion: it appears that a word is not any sequence of word characters, but only those that begin with underscore or an alphabetic character. Anything beginning with a digit is a number, not a word, and is not forced to be a string when on the left of '=>'.

The LHS of => must be a bareword for the quoting to occur. perldata defines a bareword as "a word that has no other interpretation in the grammar".

The thing is, a bareword "will be treated as if it were a quoted string." That means => does nothing except prevent the default behaviour from being "outlawed" using use strict 'subs';.

I don't know what the wording should be, but the docs for => need fixing. Submit a patch!

It sounds like you have fallen victim to the fact that perldata gives an overly narrow defintion for "bareword" while also using the term "bareword" the much broader way that I tend to.

The thing is, a bareword "will be treated as if it were a quoted string." That means => does nothing except prevent the default behaviour from being "outlawed" using use strict 'subs';.

To be clear, => does much more than prevent strict.pm from firing. => actually does some dramatic syntax transformation:

time,           # time()
time=>          # 'time'
s,this,that,    # A substitution
s=>this=>that=> # Three strings
die if 0;       # No-op
die if , 0;     # Syntax error
die if => 0;    # die "if",0;
q=<0==1         # '<0'=1
q=>0==1         # 'q',0==1
[download]

- tye        

Thanks. I realized as much after reading the related post you made yesterday. (In fact, I thought mine and yours were part of the same thread.)

-key and +key are treated differently

$ perl -Mstrict -e 'print foo => "\n"'
foo
$ perl -Mstrict -e 'print foo::bar => "\n"'
foo::bar
$ perl -Mstrict -e 'print foo:bar => "\n"'
syntax error at -e line 1, near "foo:"
Execution of -e aborted due to compilation errors.
$ perl -Mstrict -e 'print 0foo => "\n"'
syntax error at -e line 1, near "0foo"
Execution of -e aborted due to compilation errors.
$ perl -Mstrict -e "print foo'bar => qq'\n'"
foo::bar
$ perl -Mstrict -e "print foo''bar => qq'\n'"
Bad name after foo' at -e line 1.
$ perl -Mstrict -e "print foo'baz'bar => qq'\n'"
foo::baz::bar
$
[download]