It would appear to be an end of line thing. When your perl is in its own file, a comment extends to the end of the line. You can quite happyily put a semi-colon in a comment and perl just treats it as a comment.

Now extend this to the command line...

This works as expected:

perl -e 'printf "hello\n"'

This too works, as the comment is at the end:

perl -e 'printf "hello\n"; ## Comment'

This will not print anything because although the statements are separated by a semi-colon, perl has no newline to tell it where the comment ends:

perl -e '## Comment; printf "hello\n"'

This is how I entered a comment on the command line and got it to work as expected:

perl -e '## Comment; [RETURN WAS PRESSED HERE]
printf "Hello\n"'
[download]

== fx, Infinity Is Colourless

C.

I am writing the code in BBEDIT, a text editor, and then running it from the command line as an argument to the perl executable. All of my scripts run fine if I don't have any comments before, or in, the code.

an example:

%sushi = (
inari => "tofu", 
maguro => "tuna", 
ika => "squid",
tamago => "omelet",
sake => "salmon");

@english = values(%sushi);
@japanese = keys(%sushi);

foreach (@english) {
($japanese,$english) = each(%sushi); 
print ("$english => $japanese\n");
}
[download]

runs just fine, but if I add the line:

# script to swap keys and values in a hash

to the beginning, no go. The first thing I tried, thinking it was an end of line problem, was adding a semicolon to the end of my comment lines, but still no change. If I put that same comment line, no semicolon, at the end of the script, it runs just fine.

I am using Perl 5.6.0, which is what this machine came with.

Sorry, not much experience on MAC OSX, I cant see how this could be happening..

Just wanted to tell you you can reverse a hash quite easily using 'reverse' (since a hash in list context is just a list containing key,value,key,value etc.. so:

%reversedsushi = reverse(%sushi);
[download]

;)

C.

BTW - when I posted the original query I was signed in, but it posted as anonymous. I'm actually CTP, not anonymous :)

I do have the Camel book, and the Llama book, as well as several others. I had taken a couple courses a while back and bought a few, then got rusty, so now jumping back in.

I will try what you say about the BBedit saves...been using BBedit forever, but hadn't ever needed to do that so didn't even know about it.

I am reading ahead about diagnostics and 'strict' and such, and plan to write a script to ferret out the issue. I wrote here first just to make sure there wasn't one of those answers like, "Oh yea, in OSX you gotta have the comments bit set to one...sheesh!" Know what I mean? :-)

Many thanks!

ARRRRRGGGHHHHH.

No, not pirate like...Charlie Brown like. Changed BBedit to save as UNIX files. No joy :(

I think you need to do a little more than just save as Unix. You need to convert the line endings that are already there to the Unix line endings.

I haven't had BBEdit installed since I installed Panther (OS X 10.3) but I seem to remember this was pretty easy to do.

update: Check out Line Ending Converter for a conversation on line endings (as well as some scripts to convert files in place).

Just to note, this isn't MacPerl any more, it's just ... Perl running on MacOSX. OSX is UNIX-like.

And I don't think the original poster was asking about Perl programs written in the command line, only about programs run from the command line.

It might be a line-ending problem of another kind, but I second the "please provide examples" post above.

---




($_='kkvvttuubbooppuuiiffssqqffssmmiibbddllffss')
 =~y~b-v~a-z~s;                             print
[download]

Just to note, this isn't MacPerl any more, it's just ... Perl running on MacOSX. OSX is UNIX-like.

That's what I thought, but wasn't sure; hence the Mac perl, not MacPerl. Are there editors that have \x0D as newline convention while the OSX perl is expecting \x0A? Or vice versa?

And I don't think the original poster was asking about Perl programs written in the command line, only about programs run from the command line.

It might be a line-ending problem of another kind, but I second the "please provide examples" post above.

If it is a line-ending problem, pasting an example will probably hide the evidence. Maybe stick something like

BEGIN { use Data::Dumper; $Data::Dumper::Useqq=1; open FOO, $0 or die 
+"nope: $!"
; print Dumper $_ while <FOO>; exit }
[download]

at the beginning of the source and run it and show the output. I don't remember if the Data::Dumper in 5.6.0 had the Useqq option, though.

What happens when you run the program in the perl debugger?

Are you using strict and warnings?

Try putting as the first line of your script:

#!/usr/bin/perl
[download]

A comment as the very first line of the script is sometimes used by your OS and/or by Perl to determine what interpreter to run. Normally it has to start with #!, but it's possible something weird is going on. And besides, starting off your script with a proper "shebang" line is good form. :-)

If that doesn't work, show us the output, inside <code> tags, of cat -e yourscript, or whatever the equivalent is on MacOS X. That should show any nonprinting characters, and mark the ends of lines.

If I saved the file in the old Mac format with ASCII CR's s as line separator, any comments in file would extend from the comment to EOF. But if I saved the file using ASCII LF's (\0x10) as line separator, everything worked as it should.

I believe you still have not succeeded in saving your file in unix format. Use hexdump to see exactly what line separator you have.

Not (yet) knowing how to UPDATE a comment, I reply to my own. Correction: ASCII LF is \0x0a and ASCII CR is \0x0d.

So as we all suspected it was an end of line issue. What confuses me tho is why the semicolon didn't tell the interpreter that that was the end of the line...every other line is "improperly" ended, but the interpreter doesn't seem to care. I guess not all semicolons are created equal? UPDATE - I just re-read some of the previous posts, and I guess a semicolon in a comment is not created equal as it is rendered irrelevant by being in a comment...makes sense.

What happens when you run the program in the perl debugger?
Don't know how to do that yet

Are you using strict and warnings?
Not yet, but will start. Tried warnings a couple times, but got none, so moved on.
Had planned on using the shebang as well, but had to stop until I got this worked out. That's actually what tipped me off to the problem.

If that doesn't work, show us the output, inside <code> tags, of cat -e yourscript, or whatever the equivalent is on MacOS X. That should show any nonprinting...
I ran the cat -e, and it is in fact the ^M that seems to be in there munging it up.

However, the same script when created with vi ran just fine.
yea - I thought of this last night. I started doing my homework in BBedit just because it's a tool I am very used to for other work. I'll get out my little vi pocket ref and dig in...all the vi I ever knew will probably come back to me after a few scripts.

I do plan on trying to make it work the other way too tho. Many thanks to all who came to my rescue!

CTP

As for the Perl debugger, perl -d <scriptname> should get you started. It's quite useful, I've gotten my non-Perl savvy co-workers into the habit of running it whenever they set up one of our Perl scripts. I just got into the habit of using vi because on one system, that was the only editor I had available, so I learned it (the other telling point was when the fstab on my Linux box got messed up and I had to reboot into standalone mode and edit the thing by hand and at that point, vi is all you've got :-). I played a bit with BBEdit and found that if you do a "Save As" and select "Simple Text" and "ASCII" under the options (or advanced options), the file will get saved the way you want it and run just right (discovered that after I posted the script to clean up the CR/LFs. Anyhow, good luck.
---
"Ex libris un peut de tout"