Are the modules in question .pm files being used (pulled in at compile time) or .pl files being required (at run time)?

I suspect the former. If a used file is pulled in, any mainline code (not part of a sub) actually gets run at compile time - as if it were in a BEGIN block.

This is handy for initialisation code, but if your colleague is stuffing application code in here, it will indeed happen at compile time.

On the subject of package main;, this just means that any subs will be placed in the main:: namespace, thus defensively negating any package declaration prior to calling the module. IMO this should not be done, as the module should not be dictating the namespace to its caller, (unless we are talking OO, which is a different ball game).

Also note that leaving the use or require restores the namespace to that of the calling code.

For more on this, see perldoc perlfunc sections on use and require. I also recommend buying a copy of the Camel book for your colleague.

I may be able to answer more specifically if you post some code.

hth

--rW

<indent>Are the modules in question .pm files being used (pulled in at compile time) or .pl files being required (at run time)?</indent>

Yes. These files are called script_name.pm with their first line containing package main; and every module that is being written is being called from the main script as use script_name. Now, I realize that this i s just basicly placing the namespace of each script into main:: which is fine but I am still not clear why my subs are being run at compile time.

Here is a portion of my code:

     38 my %sn;                               # temporary
     39 $sn{chanserv}        = "ChanServ";    # temporary
     40 
     41 my $_name            = $sn{chanserv};
     42 my($_source,$_data)  = (shift,shift);
     43 
     44    $_data            = &_cs_parse_raw_cmd($_data) if $_data;
     45 &_sendit($_name,$_source,$_data);
     46 
     47 # ************************************************************
     48 #  sub routines
     49 # ************************************************************
     50 
     51 # ***
     52 # _send_it(): simple wrapper for chanserv to send data
     53 # through the main socket.
     54 # ***
     55 
     56 sub _sendit
     57 {
     58     my $_channel     = shift;
     59     my $_message     = shift;
     60 
     61 #    send_data(':%s ! %s :%s', $_name,  $_channel, "I saw your
+ message: $_message");
     62     printf(":%s ! %s :%s\n", $_name,  $_channel, "I saw your m
+essage: $_message");
     63 }
     64 
     65 
     66 # ***
     67 # _cs_parse_raw_cmd() : parses out raw data from main destined
     68 # for chanserv.  This will parse the raw packet.
     69 # ***
     70 
     71 sub _cs_parse_raw_cmd
     72 {
     73      my ($cmd)        = @_ ;
     74 
     75      # Chanserv :test
     76      # ChanServ@Services.Netfrag.Com :test
     77 
     78      my ($_to_whom,
     79          $_message)   = split(/:/,$cmd);
     80 
     81      # First, make sure that we are the right one getting call
+ed
     82      unless ( /\#/ ) {
     83          if ( lc($_to_whom) !~ m/$_name/i ) {
     84              &sendit($_source,"Please let an admin know that s
+omething is broke.");
     85          }
     86      }
     87 
     88      return($_message);
     89 }
     90 
     91 1;
[download]

Now. I have just spoken to the main coder about my issue and he made it clear that what I was doing was setting a scalar by calling a function which of course I was but that this was causing the problem. He also said that I needed to simply make a bunch of functions and not worry about creating any global scalars or arrays.

Everything makes sense to me now but Im still curious what the other monks think.

_ _ _ _ _ _ _ _ _ _
- Jim
Insert clever comment here...

Now. I have just spoken to the main coder about my issue and he made it clear that what I was doing was setting a scalar by calling a function which of course I was but that this was causing the problem. He also said that I needed to simply make a bunch of functions and not worry about creating any global scalars or arrays.

Nice to see that you got to the bottom of it. Global variables are generally agreed to be bad news, and show up quite clearly with

use strict;
[download]

In fact, if you try to run code with global variables in an Apache+ModPerl configuration, it will break in strange, wonderful and inconsistent ways.

As greywolf says, use strict; is your friend.

See also this thread for a discussion on use of global variables.

One of the fundemental concepts I use when coding (whether OO-style or using my functional-fu) is high cohesion, low coupling.

What this translates to is each "interdependant" block of your should almost be able to be executed standalone, ie it takes input, does some processing and spits something out. In reality this translates to having all the infrastructure to operate standalone (eg db handles, file handles etc)

So, given this coding style, it should be relatively easy to create a test.pl to test all your methods, based on the specification.

Even before you put fingers to k/b there is a process known as "bench testing" which means, taking the technical design (while still on paper) and running data thru' it (still on paper (or virtual paper)). Once you are satisfied with the paper design, you start to cut the code.

I guess the main point is planning. You must take the time to plan good code and good systems.

Well. This is a good question that I overlooked making clear when I started this thread.

The program I am writing compiles just fine by itself. It even runs at runtime but the fact that my subs get executed at runtime when they should only be run when called is what's causing the whole script to fail.

_ _ _ _ _ _ _ _ _ _
- Jim
Insert clever comment here...

Hmm. I thought I submitted a reply to this one already but I must have overlooked the submit button a few days. Anyhoo, just so everyone knows...

I was using strict; as I always do, religiously. I learned my lesson on that one way back at the beginning of my Perl coding days. This suggestion from greywolf, btw, is always an excellent suggestion, nay, rule!

The script in question was compiling fine by itself.

Thanks for the input everyone. I just needed to make those two points clear so that at least people knew I wasn't a total tard :)

_ _ _ _ _ _ _ _ _ _
- Jim
Insert clever comment here...