From @cpansprout

On Sat Apr 23 14:27:58 2016, sprout wrote:

In these examples, 5.12.4 represents perl before the each $scalar
feature. 5.14.4 includes the feature. ./perl is 5.24-to-be, in which
the feature has been removed.

$ perl5.12.4 -e 'print each 1'
Type of arg 1 to each must be hash or array (not constant item) at -e
line 1, at EOF
Execution of -e aborted due to compilation errors.
$ perl5.14.4 -e 'print each 1'
Type of arg 1 to each must be hash (not constant item) at -e line 1,
at EOF
Execution of -e aborted due to compilation errors.
$ ./perl -e 'print each 1'
Type of arg 1 to each must be hash (not constant item) at -e line 1,
at EOF
Execution of -e aborted due to compilation errors.

Notice how the last two error messages are wrong. The last one is
wrong because arrays are also accepted.

$ perl5.12.4 -e 'print each $1'
Type of arg 1 to each must be hash or array (not scalar dereference)
at -e line 1, at EOF
Execution of -e aborted due to compilation errors.
$ perl5.14.4 -e 'print each $1'
Type of argument to each on reference must be unblessed hashref or
arrayref at -e line 1.
$ ./perl -e 'print each $1'
Experimental each on scalar is now forbidden at -e line 1.

This last message is correct, but differs from the constant case.

This may not be a problem (it is, after all, just an error message;
not something encountered much in working code), but it does make
testing a little inconvenient.

The ‘hash (not constant item)’ should really be fixed, though. Should
it be changed to the old ‘hash or array’, or should it change to match
the ‘Experimental forbidden’ message? (Take note that each(1) was
never permitted.)

I now think the old message is best in this case.

Also, since some syntax errors produce multiple error messages, for the ‘Experimental forbidden’ case I suggest we give two messages, since it will be more helpful, like this:

$ perl -e 'print each $1'
Experimental each on scalar is now forbidden at -e line 1, at EOF
Type of arg 1 to each must be hash or array (not scalar dereference) at -e line 1, at EOF
Execution of -e aborted due to compilation errors.

I hope to push patches after 5.24 if nobody objects.

(Note also that I have discovered since by previous post that the hash functions use croak, whereas the array functions use yyerror. I propose making the hash functions also use yyerror, which allows for multiple messages and provides more diagnostic information, such as ‘at EOF’, ‘near blah blah blah’, etc.)

--

Father Chrysostomos

each($scalar) error message anomalies #15291

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions