But the 100k scalars don't show up in the optree. Where are they?

I don't know why B::Concise doesn't dump the contents of array, but B::Concise does show that the array is in the optree.

$ perl -MO=Concise,-exec -e'my @a = 1..1000'
1  <0> enter 
2  <;> nextstate(main 1 -e:1) v:{
3  <0> pushmark s
4  <$> const[AV ] s
              ^
              |

5  <1> rv2av lKP/1
6  <0> pushmark s
7  <0> padav[@a:1,2] lRM*/LVINTRO
8  <2> aassign[t4] vKS/COMMON
9  <@> leave[1 ref] vKP/REFC
-e syntax OK
[download]

I know that the array is there, but that doesn't convince me that those 100k scalars are built at compile time. There might just as well be a magic initializer attached to it that generates them at runtime, or at closure cloning time or whenever.

No matter how big I chose the number in the range, I can't see increased memory consumption at compile time, which is why i suspect they aren't created at compile time.

Perl 6 - second systems done right

It shows up as const AV in the opcode tree.

You can prove the list is flattened once using

I see the 1 & 2, and 4 & 8 as consts, but I'm also seeing what look to me like two loops. Perhaps you can interpret it better?:

c:\test>perl -MO=Concise
for (1..2) {
    for (4..8) {
       say $_++;
    }
 }
^Z
t  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter ->2
2     <;> nextstate(main 4 -:1) v:{ ->3
s     <2> leaveloop vK/2 ->t
7        <{> enteriter(next->p last->s redo->8) lKS/8 ->q
-           <0> ex-pushmark s ->3
-           <1> ex-list lK ->6
3              <0> pushmark s ->4
4              <$> const[IV 1] s ->5
5              <$> const[IV 2] s ->6
6           <#> gv[*_] s ->7
-        <1> null vK/1 ->s
r           <|> and(other->8) vK/1 ->s
q              <0> iter s ->r
-              <@> lineseq vKP ->-
8                 <;> nextstate(main 2 -:2) v:{ ->9
o                 <2> leaveloop vK/2 ->p
d                    <{> enteriter(next->l last->o redo->e) lKS/8 ->m
-                       <0> ex-pushmark s ->9
-                       <1> ex-list lK ->c
9                          <0> pushmark s ->a
a                          <$> const[IV 4] s ->b
b                          <$> const[IV 8] s ->c
c                       <#> gv[*_] s ->d
-                    <1> null vK/1 ->o
n                       <|> and(other->e) vK/1 ->o
m                          <0> iter s ->n
-                          <@> lineseq vKP ->-
e                             <;> nextstate(main 1 -:3) v:{ ->f
k                             <1> preinc[t7] vK/1 ->l
j                                <1> entersub[t6] sKRMS/NO(),TARG ->k
f                                   <0> pushmark s ->g
-                                   <1> ex-rv2sv sKM/1 ->h
g                                      <#> gvsv[*_] s ->h
h                                   <$> method_named[PV "say"] s ->i
i                                   <1> rv2cv /NO() ->j
l                             <0> unstack v ->m
p                 <0> unstack v ->q
- syntax OK
[download]

Sorry, there was a set of parens missing. (Fixed.) I was about to reply to your post that for (CONST..CONST) is a counting loop rather than a loop iterating over the result of a range operator. There is no range operator, so the range operator cannot be flattened. Try

perl -e"<>; exit; @a = 1..100000;"
perl -e"<>; exit; for ((), 1..100000) {}"
[download]

Results:

perl -e"<>; exit; for ((),1..   1000) {}    1,252K
perl -e"<>; exit; for ((),1..  10000) {}    1,784K
perl -e"<>; exit; for ((),1.. 100000) {}    6,028K
perl -e"<>; exit; for ((),1..1000000) {}   45,212K
[download]

Update: Added results.

You are forcing the issue there with the otherwise pointless parens.

---

With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'

Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.

"Science is about questioning the status quo. Questioning authority".

In the absence of evidence, opinion is indistinguishable from prejudice.

The start of some sanity?