enum A <Code> "P6opaque: no such attribute '$!signature'"

[p6rt]

Migrated from rt.perl.org#123457 (status was 'resolved')

Searchable as RT123457$

[p6rt]

From @raiph

See http://irclog.perlgeek.de/perl6/2014-12-19#i_9827480&#8203;:

naddiseo m​: enum A <Block>;subset B;
camelia rakudo-moar 30edf7​: OUTPUT«===SORRY!===␤P6opaque​: no such attribute
'$!signature'␤»
naddiseo Yeah.. what's that about?
I know it's something to do with "Block"
but that error is very uninformative.

raiph m​: subset B
camelia rakudo-moar 30edf7​: ( no output )
raiph m​: enum A <Block>
camelia rakudo-moar 30edf7​: ( no output )

raiph m​: enum A <Code>
camelia rakudo-moar 30edf7​: OUTPUT«===SORRY!===␤P6opaque​: no such attribute
'$!signature'␤»

Fwiw #​123407 gives a similar error.

--
raiph

[p6rt]

From @jnthn

On Thu Dec 18 19​:01​:29 2014, raiph wrote​:

See http://irclog.perlgeek.de/perl6/2014-12-19#i_9827480&#8203;:

naddiseo m​: enum A <Block>;subset B;
camelia rakudo-moar 30edf7​: OUTPUT«===SORRY!===␤P6opaque​: no such attribute
'$!signature'␤»
naddiseo Yeah.. what's that about?
I know it's something to do with "Block"
but that error is very uninformative.

raiph m​: subset B
camelia rakudo-moar 30edf7​: ( no output )
raiph m​: enum A <Block>
camelia rakudo-moar 30edf7​: ( no output )

raiph m​: enum A <Code>
camelia rakudo-moar 30edf7​: OUTPUT«===SORRY!===␤P6opaque​: no such attribute
'$!signature'␤»

Fwiw #​123407 gives a similar error.

Code, Block, etc. are the names of types used to construct code objects, and these are currently resolved using normal lexical scoping rules. Therefore, hiding them with other symbols that don't uphold the expected contract with the compiler is going to lead to things like this.

The issue first needs a decision at design level​: how should such types be located? One option is to only ever look for them in the currently effective SETTING, for example.

[p6rt]

The RT System itself - Status changed from 'new' to 'open'

[p6rt]

From @AlexDaniel

This was fixed in (2015-12-13) rakudo/rakudo@3c81e33

W4anD0eR96++ added tests in Raku/roast@89a96af777

Closing.

On 2015-01-08 12​:47​:58, jnthn@​jnthn.net wrote​:

On Thu Dec 18 19​:01​:29 2014, raiph wrote​:

See http://irclog.perlgeek.de/perl6/2014-12-19#i_9827480&#8203;:

naddiseo m​: enum A <Block>;subset B;
camelia rakudo-moar 30edf7​: OUTPUT«===SORRY!===␤P6opaque​: no such
attribute
'$!signature'␤»
naddiseo Yeah.. what's that about?
I know it's something to do with "Block"
but that error is very uninformative.

raiph m​: subset B
camelia rakudo-moar 30edf7​: ( no output )
raiph m​: enum A <Block>
camelia rakudo-moar 30edf7​: ( no output )

raiph m​: enum A <Code>
camelia rakudo-moar 30edf7​: OUTPUT«===SORRY!===␤P6opaque​: no such
attribute
'$!signature'␤»

Fwiw #​123407 gives a similar error.

Code, Block, etc. are the names of types used to construct code
objects, and these are currently resolved using normal lexical scoping
rules. Therefore, hiding them with other symbols that don't uphold the
expected contract with the compiler is going to lead to things like
this.

The issue first needs a decision at design level​: how should such
types be located? One option is to only ever look for them in the
currently effective SETTING, for example.

[p6rt]

@AlexDaniel - Status changed from 'open' to 'resolved'