How could it be an equality of manually created parse tree and the result of parse
operation:
q)t:([]date:.z.d+til 5;ccy:5#`EUR`CAN`AUS;fx:5?(1.0 0.1))
q)@[;`date;1+]t
date ccy fx
------------------
2020.11.01 EUR 0.1
2020.11.02 CAN 1
2020.11.03 AUS 0.1
2020.11.04 EUR 1
2020.11.05 CAN 0.1
q)parse"@[;`date;1+]t"
(@;::;,`date;(+;1))
`t
q)eval parse"@[;`date;1+]t"
date ccy fx
------------------
2020.11.01 EUR 0.1
2020.11.02 CAN 1
2020.11.03 AUS 0.1
2020.11.04 EUR 1
2020.11.05 CAN 0.1
q)((@;::;(),`date;(+;1));`t) ~ parse"@[;`date;1+]t"
1b
q)eval ((@;::;(),`date;(+;1));`t)
'date
[0] eval ((@;::;(),`date;(+;1));`t)
^
and the inability to evaluate the same expression via eval
?
I think I'm missing something, but what?
Upd: Is this approach correct?:
eval over ((@;;(),`date;(+;1));`t)
The placeholder in the projection just happens to have type 101h (same as identity) but internally they are different
q)type(parse"+[;2]")1
101h
q)type(parse"+[::;2]")1
101h
q)
q)null(parse"+[::;2]")1
1b
q)null(parse"+[;2]")1
0b
In your case of a single elided item, I would resolve the projection using value
or apply & then evaluate the parse tree
q)value((@;;(),`date;(+;1));`t)
@
`t
,`date
(+;1)
q)
q)(@).((@;;(),`date;(+;1));`t)
@
`t
,`date
(+;1)