Looking at your notes. Things are a little different in J8.07.
You give as examples the sentences
(;:'foo bar') ,each <'extraboxhere'
(;:'foo bar') ,L:0 'noextraboxhere'
which produce the same result.
Before J8.07, these sentences did exactly the same thing. L:0 added the
boxing to (y) and called the loop for each. L:0 is a smidgen slower on
large arguments because it calculates the level of each argument for
every cell.
In J8.07, the loop for (each) uses the new result-assembly code, which
could have benefits if the result of this sentence were being used as
input to (;) or to any f@>, but has no special effect in this case.
The loop for L:0 goes through the same code as in pre-8.07, except that
the boxing of y is avoided.
General rule: if the result of u&.> is passed to another verb though @
or &, prefer &.> to L:0 .
Henry Rich
On 7/27/2018 10:21 PM, [email protected] wrote:
HHR> I think negative level is wrongly defined.
Just a few weeks ago, i dug down into L./L:/S:/{:: level counting
myself, incl. negative levels specs. (Raul's wonderful "convert/json"
add-on made me do it. I blogged a few notes in
https://www.gaertner.de/~neitzel/nb/archives/2018/06/16/martin_groks_the_l_level_at/index.html
)
I'm interested in reading your proposed changes but it will
take me at least a week to get to it. More likely two weeks.
Martin
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
---
This email has been checked for viruses by AVG.
https://www.avg.com
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
