This text is in POD format. You should be able to read it with the perldoc command, or any other POD reader.

NAME

Inside - Find out what's inside your Perl installation

DESCRIPTION

This program will try to report which Perl modules are available on your machine, along with some other useful information. Although it's especially made to be helpful to CGI programmers, it may be of use to other Perl users as well.

Note that I've done more than a few weird things in this code in order to make it work in some odd surroundings. The right thing to do in general is to fix the broken environments, rather than to work around them. But since the purpose of this program is to diagnose some of those broken environments, I'm breaking the rules. In short: Don't Do As This Code Does! Use the accepted techniques, instead.

You should be able to run this program on nearly any machine which has Perl, either as a CGI program or stand-alone, although not under Apache/mod_perl's non-CGI environments, like Apache::Registry or Apache::PerlRun. (The ordinary Apache CGI environment is fine, whether mod_perl is installed or not.) The only(?) thing which should need changing in the program text is the location of perl in the #!-line, the first line of the program. There's also a small Configuration section near the top of the source, if you really need to have something to fiddle with.

Of course, if you're installing this program on a webserver, your local expert may need to help you to get it running. Don't ask me to do it! :-)

FAQ

  • I can't get it to work! What's wrong?

    This program tries hard to work in any normal environment, but it may be damaged or misinstalled. If you can run it in a shell (as opposed to running it as a CGI program) you may get more information about what is happening. Check the perldiag manpage for the meaning of any diagnostic messages from perl.

    If you get an error about an "Illegal character", you probably didn't use text mode ("ASCII mode") to transfer the source from one machine to another. Try again, see the perldiag manpage for more help, or ask your local expert.

    If the error message says that "Setting locale failed", check what the perllocale manpage tells you to do to fix your setup.

    If it seems to run and produces no output from the command line, check that you didn't run the configuration program 'inside.PL' rather than the real program 'inside'.

    If the error message in MacPerl complains that it "Can't emulate -{symbol} on #! line", or if the program mysteriously doesn't run at all, you probably didn't use text mode ("ASCII mode") to transfer the source to the Mac. Try again or ask your local expert.

    When you're having trouble with a CGI program in Perl, here's a handy troubleshooting guide to get you back on track.

        http://www.smithrenaud.com/public/troubleshooting_CGI.html

    If you're running the most recent version of this program and you're still stuck after using the CGI troubleshooting guide (if appropriate), working with your local expert, and thinking about it overnight, then you may try asking about it in the newsgroup comp.lang.perl.modules or writing to me. Don't send me this entire program or its output unless I ask for it! If you've got something long you wish for me to see, put it on the web and send me just the URL.

  • Where can I get the most recent version of this program?

    It should be available on CPAN.

        http://www.cpan.org/authors/id/P/PH/PHOENIX/
  • Why doesn't this program work with Apache/mod_perl?

    It does. But this is a CGI program, and Apache::Registry and similar modules don't really use CGI. They're a little different, so as to give certain benefits to some programs. This program couldn't use any of those benefits, even if it could be made compatible with those modules. In particular, it wouldn't run any faster, since nearly all of its time is spent in doing I/O. And are you going to call this program hundreds of times every second? I hope not!

    Simply run this as a normal CGI program, and Apache/mod_perl will be happy with it. If you're not sure how to do that on your machine, check with your local expert.

  • Why did you write this program?

    There are other solutions to this problem. I didn't like them. Finding the installed modules is actually a complex problem. Most proposed solutions have a number of false positives or false negatives (this one has both; see the rest of this FAQ for details). Some proposed solutions use obfuscated or incorrect code, or don't work on the web with all standard web servers and browsers. This program also has the advantage of this FAQ and (I hope) clear diagnostic messages about problems it may encounter.

  • Why can't I use module ____? This program says it's there.

    This may be a "false positive".

    This program can't tell whether a module is properly installed. (The only way to do that is to load and test the module. Figuring out how to test an arbitrary piece of code for proper functioning is provably impossible, so I decided not to try.)

    Of course, maybe the module is properly installed, but you're simply using it incorrectly. Stop doing that.

    A proper module should be distributed with tests which you can (and should) use before installing it. If your installed modules won't pass the tests, you should almost certainly (ask your administrator to) rebuild and reinstall that module, ensuring this time that it does pass the tests.

    One possible error is that you may have used the wrong capitalization in the use declaration. Check the module's documentation to see how to use it. Neither use Cgi; nor use cgi; will properly start up the CGI module.

  • Why doesn't this program find all of my modules?

    This may be a "false negative".

    The "missing" module is not installed (or not properly installed) in one of the search directories. Those directories are the ones from Perl's compiled-in @INC variable (possibly modified by an environment variable) and the extra directories whose names are included in the source of this program.

    You probably want to add your private module/library directory to the @extra_dirs list, in the Configuration section of this program. Also see what the Perl FAQ says about keeping your own module/library directory.

  • How can I include my own module/library directory?

    Add them to the @extra_dirs list, in the Configuration section. There is a similar question in the Perl FAQ, which is worth reading.

  • Can't I specify search directories with a web form?

    No. For security reasons, the directory list must be hard-coded.

  • But I want my users to be able to specify their own directories!

    Why? They can look in their own directories any time they want, can't they? If they want to install their own copy of this program, they can configure it to their needs.

    If you still want this, you probably don't understand the security implications.

  • How can I find out about module dependencies?

    Determining which modules are needed by which other modules is far beyond the scope of this program.

    There's no way to find module dependencies which always works. But any good module should check for its own dependencies at installation time. If you use the CPAN module to install and upgrade modules, it can help you with this.

    Similarly, if a program needs a module which isn't supplied with perl, this should be made clear in the program's README file, or equivalent.

  • How can I automate installing a bunch of modules?

    Some folks want to find out which modules are installed so as to automate installing those again on a new system, or with a new version of Perl. This program isn't intended to help with this. See the CPAN module's autobundle function, instead.

  • I copied some of your code to my own program, but...

    Don't do that! I break lots of rules in this program, because I have good reasons and I know what I'm doing. You don't have good reasons, and you don't know what you're doing. :-)

  • Why does the version number for module ____ come out wrong?

    For the same reason that so many version numbers are "unknown".

  • Why do so many modules show the version number as "unknown"?

    There are several possible reasons. But if you don't have a warning that gives another reason, it may be because the module author hasn't included the version number in the standard format. See the docs for ExtUtils::MakeMaker, in the section on VERSION_FROM. But (for technical reasons) this code can't be as smart as ExtUtils::MakeMaker, so it will sometimes get the version number wrong or not get it at all.

    Technical note: Okay, if you must know. ExtUtils::MakeMaker actually runs some of each module's code in order to determine its version. That could be a security hole, if the module might contain rogue code. I'm not going to take the chance. Version numbers aren't that hard to find out on your own.

    If you're a module author and this program doesn't do as well as ExtUtils::MakeMaker at determining your module's version number, please cook up a fix. Preferably, to your module, rather than to this program. :-)

  • Why are so many programs listed as "Libraries"?

    Your programs are using the file extension ".pl", which means "Perl Library". On many systems, extensions for programs aren't needed and shouldn't be used. If you must have an extension on your program names, it's best to set up your system to use ".plx", which means "Perl Executable", then use that extension instead.

    This seems to be a losing battle, since ActiveState (and others?) strongly encourage the use of the wrong extension.

    If you wish to keep some of these from showing up, add a directory or file path to the @prune_dirs array (in the Configuration section). Unless you have both libraries and programs in the same directory (yet another reason for different extensions!) you can simply list directory names to exclude them and their contents. But listing a program won't hurt you any, if you know it's not a library.

  • Why does it take so long to run?

    This program may take perhaps more than a minute to run, depending upon your system's load, the number of modules installed, and so on. It's gathering a lot of information about your system! If you're installing it as a CGI program, you may be able to make it work as an NPH-program. (This is no faster, but it does produce some output sooner, for the benefit of you impatient folks.) Set it up just like any other CGI program, but make sure that the first four characters of the file's name are "nph-", adjust the URL accordingly, and it should work automatically. If it doesn't make any difference, well, then you just have to wait for the output, that's all.

  • Why does it run faster after the first time?

    Much of the overhead of running this program is I/O. Probably your system has cached the information which it read off of your filesystem. Try again after some time, and it will be like the first time again.

  • Why do the "modules found" messages at the top always differ?

    There are some things man was not meant to know.

  • Is there any way to configure this program to ____?

    Sometimes folks want to turn part of the output on or off. Maybe they want to put their own URL into the output. Maybe they want to change something else.

    Well, you probably can't. This program is more like a stethoscope than an ultrasound machine. It doesn't have a lot of dials and knobs.

    See the intentionally-small "Configuration" section of the program, though, if you really wish to tweak something.

    Please, see the license and disclaimer before you change any code.

  • Why doesn't it find modules relative to the current directory?

    When it's run over the web (that is, as a CGI program) this program will ignore relative paths. (These are directory paths which don't start with a slash, on Unix. On other systems, there may be other kinds of relative paths, but all relative paths start from the "current working directory".)

    The current working directory is not part of the CGI specification. Since a CGI program can't rely upon it, it must always change to a non-relative directory before it can safely use a relative path. (A future version of the CGI spec may change this - but that won't help existing programs and webservers.)

  • But I use chdir before I load modules!

    You probably aren't doing that correctly. If you do know enough to do this correctly, I can instruct you no further here.

    Anyone else, just use absolute paths in @INC, and in the @extra_dirs.

  • Is this program vulnerable to any security problems?

    Every program is. See the disclaimer elsewhere in the documentation.

    One possible problem, which is beyond the scope of this program to fix, is a Denial of Service ("DoS") attack. Briefly, this program takes time to run. If someone were to set up other computers to call this program over the web as frequently as possible, your webserver could become very slow for all legitimate users. But this can happen, to some extent, with any program that remote users can run - even with your webserver itself.

    If you worry about DoS attacks using this program, simply disable it whenever you're not using it. On most Unix-type webservers, that's easy to do by using chmod(1) to set the permissions to 0. On some webservers, you may need to change the name or remove the program entirely. See your local expert for details. No, you can't enable or disable it over the web - that would defeat the purpose, wouldn't it?

  • Why don't you use warnings and 'use strict' and....?

    Because this program is purposefully written in a way which will work around various system (mis)configurations. And actually, barring quirks in some future version of perl, I am using warnings and 'use strict'. You just might not be able to see how I'm doing it. :-)

  • Why aren't you using taint checking?

    This program shouldn't need that to be secure. A program without taint checks can be secure, just as one with them can be unsafe.

    When taint checks aren't used, we may use Perl's eval() function on a string from another module. Since we first check that the module is owned and writable only by the system administrator, this doesn't open up any new security hole. (If your installed modules aren't safe, though, it opens up that existing security hole. :-) That is to say, this technique is no more insecure in general than using the modules installed on your system.

    If you wish to use taint checking, it can be enabled in the usual way, by adding the '-T' option to the $# line at the top of the program. Just know that when taint checks are enabled, you may not be able to determine the version numbers of some modules. Rarely, you may get incorrect version numbers from a few modules.

  • Why doesn't this work with perl4?

    Do you remember when O. J. Simpson was known primarily as an ex-football player who made TV commercials? Perl 4 is older than that. Give it up. It's dead. Besides, such old perl can't use modules anyway!

COPYRIGHT, DISCLAIMER, AND LICENSE

Copyright (C) 2000 by Tom Phoenix <rootbeer@redcat.com>.

THIS PROGRAM IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

This program is free software; you can distribute it under the same terms as Perl itself. I don't recommend modifying it or distributing variant versions. In fact, I discourage modifying it, unless you're sure you know what you're doing. And if you do modify something, make sure that you've clearly labeled whatever you've done. On the other hand, if you come up with a cool or useful modification, let me know. And don't forget to periodically check CPAN for updates.

    http://www.cpan.org/authors/id/P/PH/PHOENIX/

Be cautious that, if you modify this code in any way, you do not introduce security holes. Although I have, to the best of my knowledge and ability, made this program as safe as is practicable, it may have flaws which could cause undesirable effects. Still, I don't think it's too bad: I run it myself.

AUTHOR

Tom Phoenix <rootbeer@redcat.com> with plenty of help from other folks, including (in no particular order) "Tolkin, Steve" <Steve.Tolkin@fmr.com>, Mark Lybrand <markyesme@home.com>, Eric Cholet <cholet@logilune.com>, Drew Simonis <care227@attglobal.net>, Tim Conrow <tim@ipac.caltech.edu>, Richard Martin Woodward <RichardWoodward@hotmail.com>, JohnShep <john@princenaseem.com>, Mike Solomon <mike.solomon@eps.ltd.uk>, Anno Siegel <anno4000@lublin.zrz.TU-Berlin.DE>, Randall Woodman <rwoodman@verio.net>, Ken MacFarlane <ksm@dca.net>, Philip Newton <pne@cpan.org>, and anyone whose name I've accidentally omitted. It wouldn't have been possible without all this help.