$sn = $entry3;
    $$sn = new DEVICE;
[download]

    > I think the error is bogus, and that it somehow doesn't
    > like the anon reference thingie.
[download]

Try this:

    $entry->[3] = new DEVICE;
[download]

In addition: you haven't shown us all of your code, but if you're just looping through data and creating a new DEVICE for each line of the data, then adding that to an array... are you by any chance using an incremented loop counter to hold your array index? :)

If so, you should try this instead:

    push @$entry, new DEVICE;
[download]

Ah, sorry for the confusion, I'll try to clarify. I'll even post all of my code (just don't laugh).

You are right, it is not a reference or meant to be. I'm trying to create an instance of the struct anonymously. The value stored in the scalar $sn (which was in the third element of the array @entry) is going to be the name of the new struct. If that value were K123456, then $K123456 would be the name of the struct. I know this sort of anonymous calling is valid, I've used it before.

The actual names will be stored in a hash as keys. Their value pair is simply a count of how many times they are listed in the log files. In the end, when I want to tabulate and output the data, I will call each key out of the hash (foreach my $key (keys(%Unique_SN))) and then call the DEVICE struct associated with it ($$key).

That much I have done before, but it was with referenced arrays and stuff, never with a struct.

Anyway, here is my code, or what I have of it so far. It is in an early testing stage, so I don't actually do anything with the data, yet.

#!c:\perl\bin\

no strict;                        
use File::DosGlob 'glob';
use Class::Struct;
use Getopt::Std;
use vars qw($opt_a);

getopts('a:');

my $logPath = "\\\\" . $opt_a . "\\d\$\\iims\\history";

print "Searching $logPath\n";

struct ( DEVICE => { datetime => '$', devicetype => '$', zip => '$', v
+ersion => '$'});

my %Unique_DeviceType;
my %Unique_SN;

foreach my $logfile (glob("$logPath\\*")){
    open (LOG, "<$logfile") or die "Can't open $logfile:  $!";
    while(<LOG>){
        chomp;                    
        s/^\s+//;                
        s/\s+$//;                
        s/\s*,\s*/,/g;                
        s/\[|\]//g;                
        next unless length;            
        my @entry = split(/,/, $_);     #split entry:  Date Time Devic
+eType SN ZIP Version

        my @date = split(/:/, $entry[0]);     #create datetime  YYYYMM
+DDHHMMSS
        my @time = split(/:/, $entry[1]);
        my $datetime = $date[2] . $date[1] . $date[0] . $time[0] . $ti
+me[1] . $time[2];

        my $sn = $entry[3];

        $Unique_DeviceType{$entry[2]}++;     #Add entry to unique list
+ of Device Types
        $Unique_SN{$sn}++;     #Add entry to unique list of SNs
        
        if ($Unique_SN{$sn} == 1) {     #If this is first occurance
            $$sn = new DEVICE;
            $$sn->datetime('$datetime');
            $$sn->devicetype('$entry[2]');
            $$sn->zip('$entry[4]');
            $$sn->version('$entry[5]');
        }
        elsif ($datetime > $$sn->datetime) {
            $$sn->datetime('$datetime');
            $$sn->zip('$entry[4]');
            $$sn->version('$entry[5]');
        }
    }        
}

foreach my $key (keys(%Unique_SN)){
    print $key . "\n";
}
[download]

Ah, so basically, you're trying to assign the DEVICE to a symbolically referenced scalar, whose name is coming from $entry[3]. I would recommend against doing this, and for using a hash instead, to store these DEVICEs. Like this:


$devices{$sn} = new DEVICE;
$devices{$sn}->datetime{$datetime};
# etc.
[download]

If you rearranged your data slightly, you could get rid of one hash.


# I've snipped out what I consider to be the important bits...
# Warning: This is untested code!

# First, store the sn count inside the DEVICE, since we have one DEVIC
+E per $sn
struct ( DEVICE => { datetime => '$', devicetype => '$', zip => '$', v
+ersion => '$', sn_count => '$'});

# Get the $sn and other fields here

# If this is the first time we see $sn, populate the DEVICE
if (!exists $SNs{$sn}) {
    $SNs{$sn} = new DEVICE;
    $SNs{$sn}->datetime('$datetime');
# etc. 
}

# update the sn_count for this $sn.
$SNs{$sn}->sn_count($SNs{$sn}->sn_count + 1);

foreach my $key (keys %SNs) {
    print "$key\n";
}
[download]

What is your motivation for storing things in symbolically referenced scalars instead of hash elements?

Alan

$entry[3] = new DEVICE;
[download]

If you explain more specifically what you're trying to do, or how you're trying to structure your data storage, it would be a lot easier to help.

Ah, I see what you are wanting to do. I will try it. You are right, it will save a hash, and save the anon namespaces, since they will now be referenced from the hash's values.

I will try.

Thanks,
Travis

Whooohooo!! It is working.

For those of you who are keeping score, here is the final code, and what changes had to be made.

First, $foo->bar('$blah') puts "$blah" in bar, not the value of $blah. Whoops, stupid mistake.

%Unique_SN holds the unique serial number as the key and a pointer to the DEVICE struct associated with it as its value.

I can now 'use strict' agian, and I don't have to use the itermediate $sn. Now to finish this thing. I might be able to dump %Unique_DeviceType, but I'm holding on to it for now.

Thanks all. I think the orignal was not working because of the $$sn->datetime($datetime) line, not because of the $$sn = new DEVICE line. The failure was, I think, in not dereferencing. It didn't care about the anon $$sn, but $$sn was literally holding a memory address.

Wait... I think I may have just confused myself. Anyway, it is working, so no need to worry with it. I like this way better.

#!c:\perl\bin\

use strict;                            #must find way to use deref in 
+my statement
use File::DosGlob 'glob';
use Class::Struct;
use Getopt::Std;
use vars qw($opt_a);

getopts('a:');

my $logPath = "\\\\" . $opt_a . "\\d\$\\iims\\history";

print "Searching $logPath\n";

struct ( DEVICE => { datetime => '$', devicetype => '$', zip => '$', v
+ersion => '$'});

my %Unique_DeviceType;
my %Unique_SN;

foreach my $logfile (glob("$logPath\\*")){
    open (LOG, "<$logfile") or die "Can't open $logfile:  $!";
    while(<LOG>){
        chomp;                                                    #pul
+l off newline
        s/^\s+//;                                                #pull
+ out leading whitespace
        s/\s+$//;                                                #pull
+ out trailing whitespace
        s/\s*,\s*/,/g;                                            #pul
+l out whitespace around commas
        s/\[|\]//g;                                                #re
+move [ ] around date time stamp
        next unless length;                                        #mo
+ve to the next line unless there is something left in this one
        my @entry = split(/,/, $_);                                #sp
+lit entry:  Date Time DeviceType SN ZIP Version

        my @date = split(/:/, $entry[0]);                        #crea
+te datetime  YYYYMMDDHHMMSS
        my @time = split(/:/, $entry[1]);
        my $datetime = $date[2] . $date[1] . $date[0] . $time[0] . $ti
+me[1] . $time[2];

        $Unique_DeviceType{$entry[2]}++;                        #Add e
+ntry to unique list of Device Types

        if (!exists $Unique_SN{$entry[3]}) {                    #If th
+is is first occurance
            $Unique_SN{$entry[3]} = new DEVICE;                    #Ad
+d entry to unique list of SNs
            $Unique_SN{$entry[3]}->datetime($datetime);
            $Unique_SN{$entry[3]}->devicetype($entry[2]);
            $Unique_SN{$entry[3]}->zip($entry[4]);
            $Unique_SN{$entry[3]}->version($entry[5]);
        }
        elsif ($datetime > $Unique_SN{$entry[3]}->datetime) {    #If t
+his occurance is newer
            $Unique_SN{$entry[3]}->datetime($datetime);
            $Unique_SN{$entry[3]}->zip($entry[4]);
            $Unique_SN{$entry[3]}->version($entry[5]);
        }
    }        
}

foreach my $key (keys(%Unique_SN)){
    print $key . "\n";
}
[download]

You wrote:
> Thanks all. I think the orignal was not working because of the $$sn->datetime($datetime) line,
> not because of the $$sn = new DEVICE line. The failure was, I think, in not dereferencing. It
> didn't care about the anon $$sn, but $$sn was literally holding a memory address.

The -> operator does the dereferencing, but I think
$$sn->datetime($datetime)

actually means
$($sn->datetime($datetime))

and not
($$sn)->datetime($datetime)

So, it was trying to dereference the wrong thing.

Glad you got it working! Have fun.

Alan