1Module::Install::PhilosUospehry(C3o)ntributed Perl DocumMeondtualtei:o:nInstall::Philosophy(3)
2
6 Module::Install::Philosophy - The concepts behind Module::Install
7
9 This document describes the personal philosophy behind the creation of
10 CPAN::MakeMaker (the predecessor of Module::Install). The views
11 expressed here belong to Brian Ingerson; if they are not of interest to
12 you, you can safely ignore this document.
13
15 I say to you today, my friends, that in spite of the difficulties and
16 frustrations of the moment, I still have a dream. It is a dream deeply
17 rooted in the Perl Module dream.
18 I have a dream that one day this community will rise up and live out
20 the true meaning of its creed: "We hold these truths to be self-evi‐
21 dent: that all Perl authors are created equal."
22 I have a dream that one day even the state of the "CGI::" namespace, a
24 desert state, sweltering with the heat of injustice and oppression,
25 will be transformed into an oasis of freedom and justice.
26 I have a dream that my four modules will one day live in an archive
28 where they will not be judged by the number of their prerequisites but
29 by the content of their source code.
30 I have a dream today.
32
34 The above is obviously a mutation of the monumental speech by great
35 Martin Luther King (<http://web66.coled.umn.edu/new/MLK/MLK.html>).
36 While the contexts are vastly different, I feel that there are some
37 serious parallelisms.
38 The CPAN has become a place that is not free of injustice. This situa‐
40 tion has arisen not out of directed oppression, but from a failure of
41 our community to keep its tools sharp. It is the culmination of many
42 small decisions made in the name of practicality. This is a sad state
43 for an institution that was created to allow all interested people to
44 contribute equally to the best of their ability.
45 This assertion is rooted in my personal experience as an author. When I
47 created my first Perl module, Inline.pm, I knew that I had done some‐
48 thing important. But how was I to make a dent in vast Perl community?
49 As a complete unknown in the Perl community, my voice did not travel
51 far. I repeatedly tried to get even an acknowledgment from the gurus
52 familiar with XS. No success. I resorted to sending messages with
53 ridiculous subjects to "modules@perl.org".
54 (<http://www.xray.mpe.mpg.de/mailing-lists/mod‐
55 ules/2000-08/msg00078.html>) No response. Through sheer determination
56 and shameless self-promotion I eventually got the word out, and I hope
57 the world is a slightly better place for it.
58 Since then, Inline has won awards and I have had the privilege to meet
60 almost all of Perl's finest. But I still remember the pain of starting
61 out, and want to help invite more people into this wonderful world.
62 One thing I have learned from experience is that the Perl community
64 (and throw in the Python and Ruby people as well) is a small drop in
65 the vast ocean of programming. It's a giant pot of Java out there; and
66 a sea of C. Perl may not be the biggest fish, but with some care and
67 cunning we could become a much bigger school.
68 These are the current problems that I see with CPAN and the core mod‐
70 ules:
71 * New Modules don't help Older Perls
73 If I were to guess what percent of all Perl5 installations were at
74 the current release level (5.8.0 in October 2002) I would say 3-5%.
75 That may even be generous. I'd say that over 40% of installations
76 might still be at 5.005 or earlier.
77 The biggest problem with adding a module to the core is that it
79 only helps a small subset of Perl users for a long long time. Worse
80 yet, a good module author will still probably avoid using the core
81 additions as prerequisites, because they want their new module to
82 work as well on 5.005 as on 5.8.
83 CPAN::MakeMaker should be able to help in this regard. For example,
85 instead of putting Inline.pm into the core for 5.9, I can now
86 effectively get it into the core for every version of Perl that
87 Inline supports.
88 * Author Exclusiveness
90 Not just anybody can get a module into the core. It seems you have
91 to know people in high places. If I were a brilliant new talent
92 with a great new module, it would have a harder time getting the
93 ear of the pumpking, then if I were, say, Damian Conway. In fact, I
94 probably wouldn't even know where to start.
95 * Reduced Competition
97 One comment I've heard from some very good Perl programmers is
98 "Everything important has already been done". Their feeling is that
99 even though a module is suboptimal, it would be a waste of time to
100 write a competing module. Who would use it instead of the one
101 already in the core?
102 When I write a competing module, I know that I have to make it at
104 least twice as good as the existing one to even get noticed. That's
105 not a bad thing, but should everybody be forced into that situa‐
106 tion?
107 For example, let's say that you have created a really useful CGI
109 script. Let's also say that it makes use of your own CGI::Special
110 module, because CGI.pm doesn't meet your needs. Even though your
111 script might be generally useful and worth sharing, the fact that
112 it requires a non-standard module can only negatively affect its
113 acceptance. Trying to get general acceptance for the superior
114 CGI::Special module will be harder still.
115 Core modules are assumed by the general public to be "Best of
117 Breed". While this may be true for some modules at some point in
118 time, it keeps talented people from attempting to "breed" something
119 better.
120 * Core Bloat
122 Every time we add a module to the core it gets bigger and bigger.
123 And we can't ever remove modules from the core, once they've been
124 added.
125 If I had my druthers, we'd remove all modules from the core that
127 weren't necessary for either running Perl or installing modules. Of
128 course, we'd need to set things up so that installing modules was
129 so easy, that it could be done on the fly if necessary. Is this
130 easily accomplishable? Nope. Is it impossible? Nope. We have the
131 best language in the world to help us do it!
132 * Maintenance Bitrot
134 Believe it or not, Perl authors can sometimes acquire a "Life
135 Beyond Perl". They get families or new hobbies or even hit by a
136 bus. (This would be a "Death Beyond Perl".) The fact is, that once
137 somebody writes a piece of code and shares it with the world, they
138 are expected to maintain it for all time.
139 That is being generous. There are others that think that once their
141 module has become popular or made it into the core, they don't need
142 to keep fixing and improving it. I have personally been guilty of
143 this sin.
144 And then there's the Damian Conway Effect. This plagues the excep‐
146 tional authors who are so innovative and prolific they simply don't
147 have time to maintain everything they have written.
148 I initially formalized these opinions at the YAPC (Yet Another Perl
150 Conference) in June 2001. Since then I have been trying to think of
151 technological solutions to fix these social problems.
152 One idea was dubbed NAPC. NAPC is CPAN backwards. It is a large system
154 of precompiled modules that can be installed on the fly, with the goal
155 of reducing the number of modules in the core. NAPC hasn't got started
156 yet. I'd still like to do it someday, but it's a big problem with a lot
157 of issues.
158 CPAN::MakeMaker (and now Module::Install) on the other hand, is simple
160 and ultimately flexible. It should work with all of the existing CPAN
161 processes without requiring any changes from them. And new features can
162 be continuously added. Even though it doesn't scratch all of my philo‐
163 sophical CPAN itches, it's a good start.
164
166 This is all just food for thought. Take it with a pinch of salt.
167
169 Brian Ingerson <INGY@cpan.org>
170
172 Copyright (c) 2002. Brian Ingerson. All rights reserved.
173 This document is free documentation; you can redistribute it and/or
175 modify it under the same terms as Perl itself.
176 See <http://www.perl.com/perl/misc/Artistic.html>
178perl v5.8.8 2007-03-05 Module::Install::Philosophy(3)