+An example of a smaller "sandbox" is that this feature can be used to
+implement per-thread working directories: Win32 already does this.
+
+See also L</"Extend PerlIO and PerlIO::Scalar">.
+
+=head2 Investigate PADTMP hash pessimisation
+
+The peephole optimier converts constants used for hash key lookups to shared
+hash key scalars. Under ithreads, something is undoing this work.
+See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
+
+=head2 Store the current pad in the OP slab allocator
+
+=for clarification
+I hope that I got that "current pad" part correct
+
+Currently we leak ops in various cases of parse failure. I suggested that we
+could solve this by always using the op slab allocator, and walking it to
+free ops. Dave comments that as some ops are already freed during optree
+creation one would have to mark which ops are freed, and not double free them
+when walking the slab. He notes that one problem with this is that for some ops
+you have to know which pad was current at the time of allocation, which does
+change. I suggested storing a pointer to the current pad in the memory allocated
+for the slab, and swapping to a new slab each time the pad changes. Dave thinks
+that this would work.
+
+=head2 repack the optree
+
+Repacking the optree after execution order is determined could allow
+removal of NULL ops, and optimal ordering of OPs with respect to cache-line
+filling. The slab allocator could be reused for this purpose. I think that
+the best way to do this is to make it an optional step just before the
+completed optree is attached to anything else, and to use the slab allocator
+unchanged, so that freeing ops is identical whether or not this step runs.
+Note that the slab allocator allocates ops downwards in memory, so one would
+have to actually "allocate" the ops in reverse-execution order to get them
+contiguous in memory in execution order.
+
+See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
+
+Note that running this copy, and then freeing all the old location ops would
+cause their slabs to be freed, which would eliminate possible memory wastage if
+the previous suggestion is implemented, and we swap slabs more frequently.
+
+=head2 eliminate incorrect line numbers in warnings
+
+This code
+
+ use warnings;
+ my $undef;
+
+ if ($undef == 3) {
+ } elsif ($undef == 0) {
+ }
+
+used to produce this output:
+
+ Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
+ Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
+
+where the line of the second warning was misreported - it should be line 5.
+Rafael fixed this - the problem arose because there was no nextstate OP
+between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
+reports that the currently executing line is line 4. The solution was to inject
+a nextstate OPs for each C<elsif>, although it turned out that the nextstate
+OP needed to be a nulled OP, rather than a live nextstate OP, else other line
+numbers became misreported. (Jenga!)
+
+The problem is more general than C<elsif> (although the C<elsif> case is the
+most common and the most confusing). Ideally this code
+
+ use warnings;
+ my $undef;
+
+ my $a = $undef + 1;
+ my $b
+ = $undef
+ + 1;
+
+would produce this output
+
+ Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
+ Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
+
+(rather than lines 4 and 5), but this would seem to require every OP to carry
+(at least) line number information.
+
+What might work is to have an optional line number in memory just before the
+BASEOP structure, with a flag bit in the op to say whether it's present.
+Initially during compile every OP would carry its line number. Then add a late
+pass to the optimiser (potentially combined with L</repack the optree>) which
+looks at the two ops on every edge of the graph of the execution path. If
+the line number changes, flags the destination OP with this information.
+Once all paths are traced, replace every op with the flag with a
+nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
+control on to the true op. All ops would then be replaced by variants that
+do not store the line number. (Which, logically, why it would work best in
+conjunction with L</repack the optree>, as that is already copying/reallocating
+all the OPs)
+
+(Although I should note that we're not certain that doing this for the general
+case is worth it)
+
+=head2 optimize tail-calls
+
+Tail-calls present an opportunity for broadly applicable optimization;
+anywhere that C<< return foo(...) >> is called, the outer return can
+be replaced by a goto, and foo will return directly to the outer
+caller, saving (conservatively) 25% of perl's call&return cost, which
+is relatively higher than in C. The scheme language is known to do
+this heavily. B::Concise provides good insight into where this
+optimization is possible, ie anywhere entersub,leavesub op-sequence
+occurs.
+
+ perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
+
+Bottom line on this is probably a new pp_tailcall function which
+combines the code in pp_entersub, pp_leavesub. This should probably
+be done 1st in XS, and using B::Generate to patch the new OP into the
+optrees.