What does your input look like? What are the values of $i and $j?

How does your program fail?

START $i
MAY 
JUNE 
JULY
END $j
[download]

here $i and $j are variables which change. I am trying to search (with a one liner) between the pattern1 and pattern2 and prefix a variable.

perl -ne ' print if /START $i/ .. /END $j/' file.txt
[download]

EXPECTED RESULT:

START $i
$i $j MAY 
$i $j  JUNE 
$i $j  JULY
END $j
[download]

The $i and $j values change with variable

#!/usr/bin/perl
use strict;
use warnings;


while (<DATA>)
{
    process_record ($1) if /^\s*START\s*(\w+)/;  
}

sub process_record
{
   my $start_value = shift;
   my @record;
   
   my $line;
   while (defined ($line=<DATA>) and $line !~ /^\s*END/)
   {
      push @record, $line;
   }
   my ($end_value) = $line =~ /^\s*END\s+(\w+)/;
   
   @record = map{"$start_value $end_value $_"}@record;
   print @record;
   print "\n";
}

=prints
abc xyz MAY 
abc xyz JUNE 
abc xyz JULY

oue 2345 MAY 
oue 2345 JUNE 
oue 2345 JULY
=cut


__DATA__
START abc
MAY 
JUNE 
JULY
END xyz
blah
....more blah...
START oue
MAY 
JUNE 
JULY
END 2345
[download]

  # @record = map{"$start_value $end_value $_"}@record;
  # print @record;
  print "$start_value $end_value $_" for @record;
[download]

Can you advice in a one liner ?

Yes, it is possible that I could devise a "one-liner", but WHY?
I can also write code use the flip-flop operator, but WHY?
Your situation is not the ideal case for the flip-flop.

There is significant overhead in invoking Perl. There is no difference if the Perl program is expressed as one line or 20 lines.

Why are you obsessed with a "one-liner" solution? This has no performance basis.

Does my code accomplish your desired behaviour? If not, then why not?