update:

  DB<1> use List::MoreUtils "uniq"; print uniq qw(a a a a b c c a a d 
+e e e e)
abcde
[download]

Oops, adjusted:

my @b = @{ reduce { push @$a, $b if !@$a || $b ne $a->[-1]; $a } [], @
+a };
[download]

If one were to make a list version of reduce, the callback would need access to three variables: The list (say $_), the state (say $a), the current value (say $b). The problem could be solved as follows:

my @b = list_reduce { push @$_, $b if !@$_ || $b ne $_->[-1]; undef } 
+undef, @a;

my @b = list_reduce { push @$_, $b if $a || $b ne $_->[-1]; 0 } 1, @a;

my @b = list_reduce { push @$_, $b if defined($b) && $b ne $a; $b } un
+def, @a;

my @b = list_reduce { push @$_, grep defined && $_ ne $a, $b; $b } und
+ef, @a;
[download]

(@a may not start with undef for the last two to work properly.)

Update: Added everything after the first block of code.

my @b = @{ reduce { push @$a, $b if !@$a || $b ne $a->[-1]; $a } [], @
+a };
[download]

This is really clever and should be included in List::Util: https://rt.cpan.org/Ticket/Display.html?id=150773.
Note I had to disambiguate reduce by adding a '+'.

> > my @b = @{ reduce { push @$a, $b if @$a && $b ne $a->[-1]; $a } [], @a };

And I said:

> (Sure I could use $a as an array-ref for accumulation, not very elegant...)

Cheers Rolf