Opening file($ARGV) with Getopt - failing

Hi,
Why my code below fail to open and
print the file contents

when I do:

perl mycode.pl -f filename

Regards,
Edward WIJAYA
SINGAPORE

__BEGIN__
use strict;
use warnings;

use Getopt::Std;
use vars qw($f);
getopts('f:');

my	$f = $ARGV[0];
open ( INFILE, '<', $f)
or die "$0 : failed to open input file $f : $!\n";
close ( INFILE );


while ( <> )
{
print $_;
}
__END__

Hi Edward!


On Tue, 28 Sep 2004 18:51:12 +0800, Edward Wijaya
<ewijaya@singnet.com.sg> wrote:
> Hi,
> Why my code below fail to open and
> print the file contents
>
> when I do:
>
> perl mycode.pl -f filename
>
> Regards,
> Edward WIJAYA
> SINGAPORE
>
> __BEGIN__
> use strict;
> use warnings;

Good Start!  Those pragmas above are very helpful!

>
> use Getopt::Std;
> use vars qw($f);

The above is good, but is now obsolete.  The preferred method is to
use 'our' declarations
Also, the 'getopts()' function creates variables of the form 'opt_*'
where '*' is replaced with
your option name.  So, for example, you should have declared opt_f here:

our $opt_f;

> getopts('f:');
>
> my      $f = $ARGV[0];
> open ( INFILE, '<', $f)
>      or die "$0 : failed to open input file $f : $!\n";

This is good, I especially like the 'die' statement in case it
fails.  Good Job!  It is relevant
to note that opening a file to read is default, so the '<' was not
necessary.  However, it is
nice to make it obvious which way you are opening the file (read
only, write, or etc.).  I
might have written this as follows:

open INFILE, "<$opt_f" or die "$0: failed to open input file $opt_f: $!";

> close ( INFILE );
>
Why are you closing the file you just opened?  Maybe it's because
you don't understand
the diamond ('<>') operator.  The diamond operator will read the end
of you command line
and open each filename it finds there for processing.  It allows you
to write a Perl script
that acts like any other UNIX process (e.g. cat, grep, etc ... ).
In your code example,
it appears as if you are trying NOT to use the diamond operator and
force your user
to input a single filename with the '-f' option.  If this is the
case, you don't want to close
your 'INFILE' above until after you've used it!  Like this:

while( <INFILE> ) {
print;
}

> while ( <> )
> {
>      print $_;

Inside this block, the '$_' variable is default and will be assigned
the next line from
the file that 'while' is processing.  Because it is default, it is
not necessary.

> }
> __END__
>
> --
> To unsubscribe, e-mail: beginners-unsubscribe@perl.org
> For additional commands, e-mail: beginners-help@perl.org
> <http://learn.perl.org/> <http://learn.perl.org/first-response>
>
>

Edward, I could write this script two ways.  The first is the way I
prefer and it doesn't use 'Getopt::Std' at all:

#!/usr/bin/perl

use warnings;
use strict;

while( <> ) {
print;
}

That code above uses the diamond operator correctly.  The diamond
('<>') operator reads the command line and processes each file name on
the command line after your command!  So, in a command called
'perl_cat.pl' with a command line like:
# perl_cat.pl foo.txt bar.txt
The diamond operator will first open foo.txt (processed in the while
loop) and print each line, then, open bar.txt and print each of it's
lines!

However, if you are really trying to use the 'Getopt::Std' module, I'd
do it like this:

#!/usr/bin/perl

use warnings;
use strict;

our $opt_f;
getopts( 'f:' );

open INFILE, "<$opt_f" or die "$0:  Can't open file $opt_f: $!";

while( <INFILE> ) {
print;
}

In the above code, unlike yours, I don't 'close' INFILE.  That's
because Perl will close it for me at the end of my code.

I hope this helps!

--Errin

Hi again, Edward!

Just so you know, you should CC the list when you reply!

On Tue, 28 Sep 2004 22:26:55 +0800, Edward Wijaya
<ewijaya@singnet.com.sg> wrote:
> Thanks Errin,
> It works just as you suggested.
> Thanks so much for your thorough
> explanation. Glad that I learnt much from it.
> 
> I need to use Getopt, as I will increase
> the number of user given options.
>
> Regards
> Edward WIJAYA
>

I'm glad I could help!!  Just wanted to mention one last thing.  Just
because you have to use Getopt::Std doesn't mean you can't ALSO use
the diamond ('<>') operator.  Let me demonstrate:

#!/usr/bin/perl

use warnings;
use strict;
use Getopt::Std;

our $opt_p;
getopts( 'p:' );

if( $opt_p ) {
print "You used the -p flag.  The value passed was $opt_p\n";
}

while( <> ) {
print;
}

The above will print out all the lines of the file found at the END of
your command line (that's the diamond operator at work), but it will
also allow you to specify some other option with a '-p'.  So, if you
have a text file called test.txt:
Test Data
More Test Data
Other Test Data

and you call the above program with this command line:
# test_options.pl test.txt

the output will be as follows:
Test Data
More Test Data
Other Test Data

if You instead use THIS command line:
# test_options.pl -p foobar test.txt

the output will be as follows:
You used the -p flag.  The value passed was foobar.
Test Data
More Test Data
Other Test Data

I hope that makes sense.  Don't forget that the diamond operator will
see more than one filename on that command line as well:

# test_options.pl -p foobar test.txt test.txt
You used the -p flag.  The value passed was foobar.
Test Data
More Test Data
Other Test Data
Test Data
More Test Data
Other Test Data

HTH
--Errin

On Tue, 28 Sep 2004, Errin Larsen wrote:

> On Tue, 28 Sep 2004 18:51:12 +0800, Edward Wijaya
> <ewijaya@singnet.com.sg> wrote:
> 
>
>   The above is good, but is now obsolete.

That is debatable.

Gather round, and listen to the story of a log handling utility
written in perl, and a frustrated user. He's not a perl programmer,
he's a Solaris symin. For him, perl is just a tool, just another
language. And the version of perl shipped with his Solaris is
5.005_03.

He is disinclined to build and install a new version of perl just
to support a single tool. After all, he doesn't have to build
new versions of awk, or new versions of C. Why should perl be any
different?

Trouble is, the author of this tool didn't agree. He had used
our variables instead of my. Which, of course, don't work in
5.005_03. The user was quite prepared to ditch the tool altogether
and find another because it didn't Just Work. I suggested that he
go through it replacing our with my and lo and behold, it worked.

The moral of this story is that even if you have the latest and
greatest perl, you shouldn't use the latest and greatest features
unless you absolutely must. Because if you do you will severely
limit who will use your code. our in particular is really only a
convenience, saving a few key strokes at most. So don't use it.
Ever.

I'll buy a crate of beer for the first person who can show me some
real-world code which absolutely requires our and can't be
re-written to use my <update>or use vars (cos I meant to say that
originally as well)</update>.

<http://www.perlmonks.com/index.pl?node_id=393423>

The latest & greatest is only worth using if:

* it really does offer you something you need
* it really does offer you something that wasn't available before
* you really don't have to worry about portability
* there really aren't downsides to using it

For most people, in most cases, one or more of these won't be true.

It's worth it to be aware of the downsides of using a modern feature
like `our`, and to be confident that it really does make more sense to
use it over some older approach. It may be that the new ways really are
better -- I'm certainly not against progressing the language -- but if a
new feature breaks otherwise good code, is it worthwhile?


--
Chris Devers

On Tue, 28 Sep 2004 15:26:08 -0400 (EDT), Chris Devers
<cdevers@pobox.com> wrote:
> On Tue, 28 Sep 2004, Errin Larsen wrote:
> 
>
> That is debatable.
>
>     Gather round, and listen to the story of a log handling utility

<<SNIP>>

> It's worth it to be aware of the downsides of using a modern feature
> like `our`, and to be confident that it really does make more sense to
> use it over some older approach. It may be that the new ways really are
> better -- I'm certainly not against progressing the language -- but if a
> new feature breaks otherwise good code, is it worthwhile?
>
> --
> Chris Devers
>

So, what was the justification for changing 'use vars' to 'our'?  Did
the developers just want to shave down the keystrokes?  Was it an
understandability (is that a word?!) issue?  Is there any (deep down,
underneath it all) internal difference between the two?  Is there a
resource to read about this issue?  (I did read the link you supplied,
but it didn't go into WHY this changed.)

The advice I was giving in this thread was based on the following
quote in my 'perldoc Getopt::Std' documentation.  (A quote from that):

Note that, if your code is running under the recommended
"use strict 'vars'" pragma, you will need to declare these
package variables with "our":

our($opt_foo, $opt_bar);


Later in the docs it DOES say that if you don't want to declare these
as global variables, 'getopts()' will accept a hash reference.
(Another quote):

For those of you who don't like additional global variables
being created, getopt() and getopts() will also accept a
hash reference as an optional second argument. Hash keys
will be x (where x is the switch name) with key values the
value of the argument or 1 if no argument is specified.

Like this:

getopts('oif:', \%opts);  # options as normal. Values in %opts

I have a feeling that that way is the "most correct" way.  That way
the values being grabbed off the command line options will be scoped
specifically where you want them to be instead of being globals.

comments?

--Errin

Edward Wijaya wrote:
> Hi,

Hello,

> Why my code below fail to open and
> print the file contents
>
> when I do:
>
> perl mycode.pl -f filename
>
>
> __BEGIN__
> use strict;
> use warnings;
>
> use Getopt::Std;
> use vars qw($f);
> getopts('f:');

getopts( 'f:' ) creates the variable $opt_f and stores the following argumen
t
'filename' in that variable and *REMOVES* those arguments from @ARGV so that
@ARGV is now empty.


> my    $f = $ARGV[0];

Since @ARGV is now empty, $f is also empty (undef).


> open ( INFILE, '<', $f)
>     or die "$0 : failed to open input file $f : $!\n";
> close ( INFILE );
>
> while ( <> )

Since @ARGV is now empty there is nothing for <> to open and read from.


> {
>     print $_;
> }   __END__


You should do it like this instead:

use strict;
use warnings;

use Getopt::Std;

getopts( 'f:', \my %opt );

open INFILE, '<', $opt{ f } or die "$0 : failed to open input file $opt{f} :
 $!";

while ( <INFILE> ) {
print;
}

close INFILE;

__END__



John
--
use Perl;
program
fulfillment

Errin Larsen wrote:
> So, what was the justification for changing 'use vars' to 'our'?

I don't know, but I suspect it's because "our" is a complement to "my". Same
syntax (no silly qw() business), same lexical scoping, etc.

You're correct. our() should be used and 'use vars' should be considered
deprecated.

> On Tue, 28 Sep 2004, Errin Larsen wrote:
> 
>
> That is debatable.
>

Please, beginners, recognize the above word, *debatable*!!  I disagree
with the suggestion and feel that in many, many cases you will be better
off targeting newer features, especially since many times programs are
not distributed.  Once you get to a point of distribution, *and* someone
requests that you backport your script then make the change, it is just
as easy to go in reverse when necessary, than to deny yourself the use
of a tool that was added for good reason.  I don't expect my opinion to
be right, especially in all situations, but if you read this post then
you *need* to read the whole page that Chris linked to so that you can
see what other's thoughts were on the subject.

>     Gather round, and listen to the story of a log handling utility
>     written in perl, and a frustrated user. He's not a perl programmer,
>     he's a Solaris symin. For him, perl is just a tool, just another
>     language. And the version of perl shipped with his Solaris is
>     5.005_03.
>
>     He is disinclined to build and install a new version of perl just
>     to support a single tool. After all, he doesn't have to build
>     new versions of awk, or new versions of C. Why should perl be any
>     different?
>

Well then to me he is disinclined to use the tool, to each his own,
"don't let the door hit you on the ass on the way out" is an acceptable
answer, and often should be given more....

>     Trouble is, the author of this tool didn't agree. He had used
>     our variables instead of my. Which, of course, don't work in
>     5.005_03. The user was quite prepared to ditch the tool altogether
>     and find another because it didn't Just Work. I suggested that he
>     go through it replacing our with my and lo and behold, it worked.
>

Of course, equating 'my' with 'our/use vars' we all know to be wrong...
don't we.

>     The moral of this story is that even if you have the latest and
>     greatest perl, you shouldn't use the latest and greatest features
>     unless you absolutely must. Because if you do you will severely
>     limit who will use your code. our in particular is really only a
>     convenience, saving a few key strokes at most. So don't use it.
>     Ever.
>

This is opinion taken out of context, and stated as a rule rather than a
suggestion, which may be a bit over the top.

>     I'll buy a crate of beer for the first person who can show me some
>     real-world code which absolutely requires our and can't be
>     re-written to use my <update>or use vars (cos I meant to say that
>     originally as well)</update>.
>
> <http://www.perlmonks.com/index.pl?node_id=393423>
>
> The latest & greatest is only worth using if:
>

I disagree completely.  Two words, "buffer overflow".  See an M$ bug
list for more.... If this were the case then we would all be writing in
assembly......

>  * it really does offer you something you need
>  * it really does offer you something that wasn't available before
>  * you really don't have to worry about portability
>  * there really aren't downsides to using it
>
> For most people, in most cases, one or more of these won't be true.
>
> It's worth it to be aware of the downsides of using a modern feature
> like `our`, and to be confident that it really does make more sense to
> use it over some older approach. It may be that the new ways really are
> better -- I'm certainly not against progressing the language -- but if a
> new feature breaks otherwise good code, is it worthwhile?
>

Definitely worth knowing about the downsides, but I would push the other
direction, is it worth not using something that benefits to prevent
something that may never happen?

>
> --
> Chris Devers
>

http://danconia.org

On Tue, 28 Sep 2004, Wiggins d Anconia wrote:

> On Tue, 28 Sep 2004, Chris Devers wrote:
> 
>
> Please, beginners, recognize the above word, *debatable*!!

<aol />

I'm not forwarding this along as a blanket assertion that 'our' is bad,
broken, and always to be avoided. None of that is true. Rather, I'm
using it as an example of how a seemingly innocuous new feature can have
unexpected & possibly hard to debug consequences "in the wild".

If you're only writing for yourself on systems that you have fulll
control over, then this particular example probably isn't a big deal.
If, on the other hand, you get a job where you have to work with the
available software, then you may be stuck.

It is worthwhile to spend a few minutes figuring out what problem a new
feature solves, reflecting on whether that problem impacts you (chances
are good that it does, but it might not), and considering whether or not
using this new construct over older methods introduces new problems (it
should not, but it might).

To just blindly accept a suggestion like this without at least a little
bit of thought is what is called "cargo culting"; it's a bad habit:

<http://catb.org/~esr/jargon/html/C/...rogramming.html>
<http://www.physics.brocku.ca/etc/ca...lt_science.html>
<http://en.wikipedia.org/wiki/Cargo_cult>

That said, here's one article (of several, surely) that contrasts the
'my' and 'our' functions:

<http://perlmonks.thepen.com/105446.html>

Be aware of what's going on here before concluding that 'our' is always
the one to use. It might be, maybe, but it might not.

That's all :-)
 
>
> This is opinion taken out of context, and stated as a rule rather than a
> suggestion, which may be a bit over the top.

Maybe so, but I thought it was striking that Jokob Neilsen has written
similar things about adopting new features in web design. Consider the
footnote for this essay:

Normally, I advise against using new Web technologies for the first
year after they have been introduced. In most cases, using anything
new is asking for trouble and will alienate users with older browsers.

Link titles are an exception to the need to wait a year. First, their
use does not hurt users with browsers that don't display link titles
[....] Second, a browser that does not understand link titles will
simply skip them. [....] The only downside is that link titles will
add approximately 0.1 seconds to the download time for a typical
Web page over a modem connection. This is a rather harsh penalty,
but worth paying because of the increased navigation usability for
those users who do get to see the link titles.

At the time of this writing, link titles will probably only be seen
by 25% of the users. Normally, this would be insufficient to employ
a new Web technology, but since link titles are extremely easy to
add to your pages and since they do not pose any danger to users
with older browsers, it makes sense to go ahead and start using link
titles. [....]

<http://www.useit.com/alertbox/980111.html>

Note the way he thinks this through. The base instinct is conservative:
most new features introduce problems bigger than the one they solve, so
he avoids them. In this case, however, the problem is broad, the risks
are slight, and the benefits are large, so he breaks the usual rule.

That, I think, is the right approach to new features in any language.

> Definitely worth knowing about the downsides, but I would push the
> other direction, is it worth not using something that benefits to
> prevent something that may never happen?

Beats me -- is it?

I don't think there's a blanket rule here, one way or the other. It
comes down to what your expectations are for your code, whether or not
it could have a life of its own after you're done with it, etc.

If nothing else, it might not be bad to start off scripts using new
features with a

require 5.6.1;

(or whatever the first version with 'our' was), with a comment saying
what construct you're using that demands the require statement.


But anyway, I'll leave it at that, as this is turning into a long
discussion with no real right or wrong sides but lots of talk... :-)



--
Chris Devers