Optimize reverse RANGE

[grr]

Description

Optimize reverse RANGE so that it doesn't produce a temporary array and "burn" memory.

I came upon an ancient perlmonks thread that suggested this would be "an easy candidate for optimization", but was surprised to find nobody had actually requested it.

Steps to Reproduce

for (reverse 0..1e7) { }

Expected behavior

No use of a temporary array and hence no memory inflation.

Perl configuration

This is perl 5, version 40, subversion 0 (v5.40.0) built for darwin-2level

[richardleach]

perl -MO=Concise -E 'for (0..1e7) { }' gives:

d  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter v ->2
2     <;> nextstate(main 2 -e:1) v:%,us,{,fea=15 ->3
c     <2> leaveloop vK/2 ->d
7        <{> enteriter(next->9 last->c redo->8) vKS/DEF ->a
-           <0> ex-pushmark s ->3
-           <1> ex-list lK ->6
3              <0> pushmark s ->4
4              <$> const[IV 0] s ->5
5              <$> const[NV 10000000] s ->6
6           <#> gv[*_] s ->7
-        <1> null vK/1 ->c
b           <|> and(other->8) vK/1 ->c
a              <0> iter s ->b
-              <@> lineseq vK ->-
8                 <0> stub v ->9
9                 <0> unstack v ->a

perl -MO=Concise -E 'for (reverse 0..1e7) { }' gives:

d  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter v ->2
2     <;> nextstate(main 2 -e:1) v:%,us,{,fea=15 ->3
c     <2> leaveloop vK/2 ->d
7        <{> enteriter(next->9 last->c redo->8) vK/DEF,REVERSED ->a
-           <0> ex-pushmark s ->3
-           <1> ex-list lKM ->6
3              <0> pushmark s ->4
-              <1> ex-reverse lKM/1 ->-
-                 <0> ex-pushmark s ->4
5                 <1> rv2av lKP/1 ->6
4                    <$> const[AV ARRAY] s ->5
6           <#> gv[*_] s ->7
-        <1> null vK/1 ->c
b           <|> and(other->8) vK/1 ->c
a              <0> iter s/REVERSED ->b
-              <@> lineseq vK ->-
8                 <0> stub v ->9
9                 <0> unstack v ->a

OP_REVERSE is not involved.

Addressing this request might involve:

• Changing the optree construction to again give 2x OP_CONST, instead of 1x OP_CONST + 1x OP_RV2AV
• Changing how OP_ENTERITER sets up the context
• Adding a CXt_LOOP_LAZYIV_REV case to OP_ITER

I've no idea exactly how easy/feasible this is.