LDP
/
old-www
1
1
Fork 0
Code Issues Pull Requests Releases Wiki Activity
old-www/HOWTO/Software-Proj-Mgmt-HOWTO/starting.html

1650 lines
46 KiB
HTML
Raw Permalink Blame History

<HTML
><HEAD
><TITLE
>Starting a Project</TITLE
><META
NAME="GENERATOR"
CONTENT="Modular DocBook HTML Stylesheet Version 1.76b+
"><LINK
REL="HOME"
TITLE="Free Software Project Management HOWTO"
HREF="index.html"><LINK
REL="PREVIOUS"
TITLE="Introduction"
HREF="intro.html"><LINK
REL="NEXT"
TITLE="Maintaining a Project: Interacting with Developers"
HREF="developers.html"></HEAD
><BODY
CLASS="SECT1"
BGCOLOR="#FFFFFF"
TEXT="#000000"
LINK="#0000FF"
VLINK="#840084"
ALINK="#0000FF"
><DIV
CLASS="NAVHEADER"
><TABLE
SUMMARY="Header navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TH
COLSPAN="3"
ALIGN="center"
>Free Software Project Management HOWTO</TH
></TR
><TR
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="bottom"
><A
HREF="intro.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="80%"
ALIGN="center"
VALIGN="bottom"
></TD
><TD
WIDTH="10%"
ALIGN="right"
VALIGN="bottom"
><A
HREF="developers.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
></TABLE
><HR
ALIGN="LEFT"
WIDTH="100%"></DIV
><DIV
CLASS="SECT1"
><H1
CLASS="SECT1"
><A
NAME="STARTING">2. Starting a Project</H1
><P
> With very little argument, the beginning is the most difficult
period in a project's life to do successful free software project
management. Laying a firm foundation will determine whether your
project flourishes or withers away and dies. It is also the subject
that is of most immediate interest to anyone reading this document
as a tutorial.
</P
><P
> Starting a project involves a dilemma that you as a developer must
try and deal with: no potential user for your program is interested
in a program that doesn't work, while the development process that
you want to employ holds involvement of users as imperative.
</P
><P
> It is in these dangerous initial moments that anyone working to
start a free software project must try and strike a balance along
these lines. One of the most important ways that someone trying to
start a project can work toward this balance is by establishing a
solid framework for the development process through some of the
suggestions mentioned in this section.
</P
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="CHOOSEPROJECT">2.1. Choosing a Project</H2
><P
> If you are reading this document, there's a good chance you
already have an idea for a project in mind. Chances are also
pretty good that it fills a perceived gap by doing something that
no other free software project does or by doing something in a way
that is unique enough to necessitate a brand new piece of
software.
</P
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="IDENTIFYIDEA">2.1.1. Identify and articulate your idea</H3
><P
> Eric S. Raymond writes about how free software projects start in
his essay, <A
HREF="http://www.tuxedo.org/~esr/writings/cathedral-bazaar/"
TARGET="_top"
><SPAN
CLASS="QUOTE"
>"The
Cathedral and the Bazaar,"</SPAN
></A
> which comes as required
reading for any free software developer. It is available online .
</P
><P
> In <SPAN
CLASS="QUOTE"
>"The Cathedral and the Bazaar,"</SPAN
> Raymond tells us
that: <SPAN
CLASS="QUOTE"
>"every good work of software starts by scratching
a developers itch."</SPAN
> Raymond's now widely accepted
hypothesis is that new free software programs are written, first
and foremost, to solve a specific problem facing the developer.
</P
><P
> If you have an idea for a program in mind, chances are good that
it targets a specific problem or <SPAN
CLASS="QUOTE"
>"itch"</SPAN
> you want to
see scratched. <EM
>This idea is the project.</EM
>
Articulate it clearly. Write it out. Describe the problem you
will attack in detail. The success of your project in tackling a
particular problem will be tied to your ability to identify that
problem clearly early on. Find out exactly what it is that you
want your project to do.
</P
><P
> Monty Manley articulates the importance of this initial step in
an essay, <SPAN
CLASS="QUOTE"
>"<A
HREF="http://news.linuxprogramming.com/news_story.php3?ltsn=2000-10-31-001-05-CD"
TARGET="_top"
>Managing
Projects the Open Source Way.</A
>"</SPAN
> As the next section
will show, there is <EM
>a lot</EM
> of work that needs
to be done before software is even ready to be coded. Manley
says, <SPAN
CLASS="QUOTE"
>"Beginning an OSS project properly means that a
developer must, first and foremost, avoid writing code too
soon!"</SPAN
>
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="EVALULATEIDEA">2.1.2. Evaluate your idea</H3
><P
> In evaluating your idea, you need to first ask yourself a few
questions. This should happen before you move any further
through this HOWTO. Ask yourself: <EM
>Is the free software
development model really the right one for your
project?</EM
>
</P
><P
> Obviously, since the program scratches your itch, you are
definitely interested in seeing it implemented in code. But,
because one hacker coding in solitude fails to qualify as a free
software development effort, you need to ask yourself a second
question: <EM
>Is anybody else interested?</EM
>
</P
><P
> Sometimes the answer is a simple <SPAN
CLASS="QUOTE"
>"no."</SPAN
> If you want
to write a set of scripts to sort <EM
>your</EM
>
<SPAN
CLASS="ACRONYM"
>MP3</SPAN
> collection on <EM
>your</EM
>
machine, <EM
>maybe</EM
> the free software development
model is not the best one to choose. However, if you want to
write a set of scripts to sort <EM
>anyone's</EM
>
<SPAN
CLASS="ACRONYM"
>MP3</SPAN
>s, a free software project might fill a
useful gap.
</P
><P
> Luckily, the Internet is a place so big and so diverse that,
chances are, there is someone, somewhere, who shares your
interests and who feels the same <SPAN
CLASS="QUOTE"
>"itch."</SPAN
> It is the
fact that there are so many people with so many similar needs and
desires that introduces the third major question: <EM
>Has
somebody already had your idea or a reasonably similar
one?</EM
>
</P
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="EVALWHERE">2.1.2.1. Finding Similar Projects</H4
><P
> There are places you can go on the web to try and answer the
question above. If you have experience with the free software
community, you are probably already familiar with many of these
sites. All of the resources listed below offer searching of
their databases:
</P
><P
> <P
></P
><DIV
CLASS="VARIABLELIST"
><DL
><DT
>freshmeat.net</DT
><DD
><P
><A
HREF="http://freshmeat.net"
TARGET="_top"
>freshmeat.net</A
>
describes itself as, <SPAN
CLASS="QUOTE"
>"the Web's largest index of Linux
and Open Source software"</SPAN
> and its reputation along
these lines is totally unparalleled and unquestioned. If you
can't find it on freshmeat, its doubtful that you (or anyone
else) will find it at all.</P
></DD
><DT
>Slashdot</DT
><DD
><P
><A
HREF="http://slashdot.org"
TARGET="_top"
>Slashdot</A
>
provides <SPAN
CLASS="QUOTE"
>"News for Nerds. Stuff that matters,"</SPAN
>
which usually includes discussion of free software, open
source, technology, and geek culture news and events. It is
not unusual for a particularly sexy development effort to be
announced here, so it is definitely worth checking.</P
></DD
><DT
>SourceForge</DT
><DD
><P
><A
HREF="http://sourceforge.net"
TARGET="_top"
>SourceForge</A
>
houses and facilitates a growing number of open source and
free software projects. It is also quickly becoming a nexus
and a necessary stop for free software
developers. SourceForge's <A
HREF="http://sourceforge.net/softwaremap/trove_list.php"
TARGET="_top"
>software
map</A
> and <A
HREF="http://sourceforge.net/new/"
TARGET="_top"
> new
release</A
> pages should be necessary stops before
embarking on a new free software project. SourceForge also
provides a <A
HREF="http://sourceforge.net/snippet/"
TARGET="_top"
>Code Snippet
Library</A
> which contains useful reusable chunks of code
in an array of languages which can come in useful in any
project.</P
></DD
><DT
>Google and Google's Linux Search</DT
><DD
><P
><A
HREF="http://www.google.com"
TARGET="_top"
>Google</A
> and
<A
HREF="http://www.google.com/linux"
TARGET="_top"
> Google's Linux
Search</A
>, provides powerful web searches that may reveal
people working on similar projects. It is not a catalog of
software or news like freshmeat or Slashdot, but it is worth
checking to make sure you aren't pouring your effort into a
redundant project.</P
></DD
></DL
></DIV
>
</P
></DIV
><DIV
CLASS="SECT4"
><H4
CLASS="SECT4"
><A
NAME="EVALHOW">2.1.2.2. Deciding to Proceed</H4
><P
> Once you have successfully charted the terrain and have an idea
about what kinds of similar free software projects exist, every
developer needs to decide whether to proceed with their own
project. It is rare that a new project seeks to accomplish a
goal that is not at all similar or related to the goal of
another project. Anyone starting a new project needs to ask
themselves: <SPAN
CLASS="QUOTE"
>"Will the new project be duplicating work done
by another project? Will the new project be competing for
developers with an existing project? Can the goals of the new
project be accomplished by adding functionality to an existing
project?"</SPAN
>
</P
><P
> If the answer to any of these questions is <SPAN
CLASS="QUOTE"
>"yes,"</SPAN
>
try to contact the developer of the existing project(s) in
question and see if he or she might be willing to collaborate
with you.
</P
><P
> For many developers this may be the single most difficult aspect
of free software project management, but it is an essential one. It is
easy to become fired up by an idea and get caught up in the
momentum and excitement of a new project. It is often extremely
difficult to do, but it is important that any free software
developer remembers that the best interests of the free software
community and the quickest way to accomplish your own project's
goals and the goals of similar projects can often be
accomplished by <EM
>not</EM
> starting a new
development effort.
</P
></DIV
></DIV
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="NAMING">2.2. Naming your project</H2
><P
> While there are plenty of projects that fail with descriptive
names and plenty that succeed without them, I think naming your
project is worth giving a bit of thought. Leslie Orchard tackles
this issue in an <A
HREF="http://www.advogato.org/article/67.html"
TARGET="_top"
>Advogato
article</A
>. His article is short and definitely worth looking
over quickly.
</P
><P
> The synopsis is that Orchard recommends you pick a name where,
after hearing the name, many users or developers will both:
</P
><P
> <P
></P
><UL
><LI
><P
>Know what the project does.</P
></LI
><LI
><P
>Remember it tomorrow.</P
></LI
></UL
>
</P
><P
> Humorously, Orchard's project, <SPAN
CLASS="QUOTE"
>"Iajitsu,"</SPAN
> does
neither. It is probably unrelated that development has effectively
frozen since the article was written.
</P
><P
> He makes a good point though. There are companies whose only job
is to make names for pieces of software. They make
<EM
>ridiculous</EM
> amount of money doing it and are
supposedly worth it. While you probably can't afford a company like
this, you can afford to learn from their existence and think a
little bit about the name you are giving your project because it
<EM
>does</EM
> matter.
</P
><P
> If there is a name you really want but it doesn't fit Orchard's
criteria, you can still go ahead. I thought <SPAN
CLASS="QUOTE"
>"gnubile"</SPAN
>
was one of the best I'd heard for a free software project ever and
I still talk about it long after I've stopped using the
program. However, if you can be flexible on the subject, listen to
Orchard's advice. It might help you.
</P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="LICENSING">2.3. Licensing your Software</H2
><P
> On one (somewhat simplistic) level, the difference between a piece
of free software and a piece of propriety software is the
license. A license helps you as the developer by protecting your
legal rights to have your software distributed under your terms
and helps demonstrate to those who wish to help you or your
project that they are encouraged to join.
</P
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="CHOOSELICENSE">2.3.1. Choosing a license</H3
><P
> Any discussion of licenses is also sure to generate at least a
small flame war as there are strong feelings that some free
software licenses are better than others. This discussion also
brings up the question of <SPAN
CLASS="QUOTE"
>"Open Source Software"</SPAN
> and
the debate over the terms <SPAN
CLASS="QUOTE"
>"Open Source Software"</SPAN
> and
<SPAN
CLASS="QUOTE"
>"Free Software"</SPAN
>. However, because I've written the
Free Software Project Management HOWTO and not the Open Source
Software Project Management HOWTO, my own allegiances in this
argument are in the open.
</P
><P
> In attempting to reach a middle ground through diplomacy without
sacrificing my own philosophy, I will recommend picking any
license that conforms to the <A
HREF="http://www.debian.org/social_contract"
TARGET="_top"
>Debian Free Software
Guidelines</A
>. Originally compiled by the Debian project
under Bruce Perens, the <SPAN
CLASS="ACRONYM"
>DFSG</SPAN
> forms the first
version of the <A
HREF="http://www.opensource.org/docs/definition_plain.html"
TARGET="_top"
>Open
Source Definition.</A
> Examples of free licenses given by the
<SPAN
CLASS="ACRONYM"
>DFSG</SPAN
> are the <SPAN
CLASS="ACRONYM"
>GPL</SPAN
>, the
<SPAN
CLASS="ACRONYM"
>BSD</SPAN
>, and the Artistic License. As ESR mentions
in his his HOWTO<A
HREF="b811.html#ESRHOWTO"
>[ESRHOWTO]</A
>, don't write your own
license if at all possible. The three licenses I mention all have
long interpretive traditions. They are also definitely free
software (and can therefore be distributed as part of Debian and
in other places that permit the transfer of free software).
</P
><P
> Conforming to the definition of free software offered by Richard
Stallman in <A
HREF="http://www.gnu.org/philosophy/free-sw.html"
TARGET="_top"
><SPAN
CLASS="QUOTE"
>"The Free
Software Definition"</SPAN
></A
>, any of these licenses will
uphold, <SPAN
CLASS="QUOTE"
>"users' freedom to run, copy, distribute, study,
change and improve the software."</SPAN
> There are plenty of
other licenses that also conform to the <SPAN
CLASS="ACRONYM"
>DFSG</SPAN
>
but sticking with a more well-known license will offer the
advantage of immediate recognition and understanding. Many
people write three or four sentences in a COPYING file and assume
that they have written a free software license--as my long
experience with the debian-legal mailing professes, this is very
often not the case.
</P
><P
> In attempting a more in-depth analysis, I agree with Karl Fogel's
description of licenses as falling into two groups: those that
are the <SPAN
CLASS="ACRONYM"
>GPL</SPAN
> and those that are not the
<SPAN
CLASS="ACRONYM"
>GPL</SPAN
>.
</P
><P
> Personally, I license all my software under the
<SPAN
CLASS="ACRONYM"
>GPL</SPAN
>. Created and protected by the Free
Software Foundation and the GNU Project, the
<SPAN
CLASS="ACRONYM"
>GPL</SPAN
> is the license for the Linux kernel,
GNOME, Emacs, and the vast majority of GNU/Linux software. It's
the obvious choice but I also believe it is a good one. Any BSD
fanatic will urge you to remember that there is a viral aspect to
the <SPAN
CLASS="ACRONYM"
>GPL</SPAN
> that prevents the mixture of
<SPAN
CLASS="ACRONYM"
>GPL</SPAN
>'ed code with non-<SPAN
CLASS="ACRONYM"
>GPL</SPAN
>'ed
code. To many people (myself included), this is a benefit, but to
some, it is a major drawback.
</P
><P
> Many people write three or four sentences in a COPYING file and
assume that they have written a free software license--as my long
experience with the debian-legal mailing professes, this is very
often not the case. It may not protect you, it may not protect
your software, and it may make things very difficult for people
that want to use your software but who pay a lot of attention to
the subtle legal points of licenses. If you are passionate about
a home-brewed license, run it by either people at <A
HREF="http://www.opensource.org"
TARGET="_top"
>OSI</A
> or the <A
HREF="mailto:debian-devel@lists.debian.org"
TARGET="_top"
>debian-legal mailing
list</A
> first protect yourself from unanticipated
side-effects of your license.
</P
><P
> The three major licenses can be found at the following locations:
</P
><P
> <P
></P
><UL
><LI
><P
><A
HREF="http://www.gnu.org/copyleft/gpl.html"
TARGET="_top"
>The GNU
General Public License</A
></P
></LI
><LI
><P
><A
HREF="http://www.debian.org/misc/bsd.license"
TARGET="_top"
>The
BSD License</A
></P
></LI
><LI
><P
><A
HREF="http://language.perl.com/misc/Artistic.html"
TARGET="_top"
>The Artistic
License</A
></P
></LI
></UL
>
</P
><P
> <EM
>In any case, please read through any license before
your release your software under it. As the primary developer,
you can't afford any license surprises.</EM
>
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="LICENSECHOOSE">2.3.2. The mechanics of licensing</H3
><P
> The text of the <SPAN
CLASS="ACRONYM"
>GPL</SPAN
> offers <A
HREF="http://www.gnu.org/copyleft/gpl.html#SEC4"
TARGET="_top"
>a good
description of the mechanics of applying a license</A
> to a
piece of software. My quick checklist for applying a license
includes:
</P
><P
> <P
></P
><UL
><LI
><P
>Make yourself or the FSF the copyright holder for the
work. In a few rare cases, you might want to make a sponsoring
organization (if it's big and powerful enough) the copyright
holder instead. Doing this is as simple as putting the name in
the blank when you modify the notice of copyright
below. Contrary to popular belief, you don't need to file with
any organization. The notice alone is enough to copyright your
work.</P
></LI
><LI
><P
>If at all possible, attach and distribute a full copy of
the license with the source and binary by including a separate
file.</P
></LI
><LI
><P
>At the top of each source file in your program, attach a
notice of copyright and include information on where the full
license can be found. The <SPAN
CLASS="ACRONYM"
>GPL</SPAN
> recommends
that each file begin with:</P
><TABLE
BORDER="1"
BGCOLOR="#E0E0E0"
WIDTH="90%"
><TR
><TD
><FONT
COLOR="#000000"
><PRE
CLASS="SCREEN"
><EM
>one line to give the program's name and an idea of what it does.</EM
>
Copyright (C) yyyy name of author
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License
as published by the Free Software Foundation; either version 2
of the License, or (at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
</PRE
></FONT
></TD
></TR
></TABLE
><P
> The <SPAN
CLASS="ACRONYM"
>GPL</SPAN
> goes on to recommend attaching
information on methods for contacting you (the author) via
email or physical mail.
</P
></LI
><LI
><P
> The <SPAN
CLASS="ACRONYM"
>GPL</SPAN
> continues and suggests that if your
program runs in an interactive mode, you should write the
program to output a notice each time it enters interactive
mode that includes a message like this one that points to full
information about the programs license:
</P
><TABLE
BORDER="1"
BGCOLOR="#E0E0E0"
WIDTH="90%"
><TR
><TD
><FONT
COLOR="#000000"
><PRE
CLASS="SCREEN"
>Gnomovision version 69, Copyright (C) year name of author
Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
type `show w'. This is free software, and you are welcome
to redistribute it under certain conditions; type `show c'
for details.
</PRE
></FONT
></TD
></TR
></TABLE
></LI
><LI
><P
>Finally, it might be helpful to include a
<SPAN
CLASS="QUOTE"
>"copyright disclaimer"</SPAN
> from an employer or a
school if you work as a programmer or if it seems like your
employer or school might be able to make an argument for
ownership of your code later on. These aren't often needed but
there are plenty of free software developers who have gotten
into trouble and wish they'd asked for one.</P
></LI
></UL
>
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="LICENSEWARNING">2.3.3. Final license warning</H3
><P
> Please, please, please, place your software under
<EM
>some</EM
> license. It may not seem important, and
to you it may not be, but licenses <EM
>are</EM
>
important. For a piece of software to be included in the Debian
GNU/Linux distribution, it must have a license that fits the
<A
HREF="http://www.debian.org/social_contract"
TARGET="_top"
>Debian Free
Software Guidelines</A
>. If your software has no license, it
can not be distributed as a package in Debian until you
re-release it under a free license. Please save yourself and
others trouble by releasing the first version of your software
with a clear license.
</P
></DIV
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="CHOOSEVERSIONING">2.4. Choosing a Method of Version Numbering</H2
><P
> <EM
>The most important thing about a system of version
numbering is that there is one.</EM
> It may seem pedantic to
emphasize this point but you'd be surprised at the number of
scripts and small programs that pop up without any version number
at all.
</P
><P
> <EM
>The second most important thing about a system of
numbering is that the numbers always go up.</EM
> Automatic
version tracking systems and people's sense of order in the
universe will fall apart if version numbers don't rise. It doesn't
<EM
>really</EM
> matter if 2.1 is a big jump and
2.0.005 is a small jump but it does matter that 2.1 is more recent
than 2.0.005.
</P
><P
> Follow these two simple rules and you will not go (too)
wrong. Beyond this, the most common technique seems to be the
<SPAN
CLASS="QUOTE"
>"major level,"</SPAN
> <SPAN
CLASS="QUOTE"
>"minor level,"</SPAN
>
<SPAN
CLASS="QUOTE"
>"patch level"</SPAN
> version numbering scheme. Whether you
are familiar with the name or not, you interact with it all the
time. The first number is the major number and it signifies major
changes or rewrites. The second number is the minor number and it
represents added or tweaked functionality on top of a largely
coherent structure. The third number is the patch number and it
usually will only refer to releases fixing bugs.
</P
><P
> The widespread use of this scheme is why I know the nature and
relative degree in the differences between a 2.4.12 release of the
Linux kernel and a 2.4.11, 2.2.12, and 1.2.12 without knowing
anything about any of the releases.
</P
><P
> You can bend or break these rules, and people do. But beware, if
you choose to, someone will get annoyed, assume you don't know,
and try and educate you, probably not nicely. I always follow this
method and I implore you to do so as well.
</P
><P
> There are several version numbering systems that are well known,
useful, and that might be worth looking into before you release
your first version.
</P
><P
></P
><DIV
CLASS="VARIABLELIST"
><DL
><DT
>Linux kernel version numbering:</DT
><DD
><P
>The Linux kernel uses a versioning system where any odd
minor version number refers to an development or testing release
and any even minor version number refers to a stable
version. Think about it for a second. Under this system, 2.1 and
2.3 kernels were and always will be development or testing
kernels and 2.0, 2.2. and 2.4 kernels are all production code
with a higher degree of stability and more testing.
</P
><P
> Whether you plan on having a split development model (as
described in <A
HREF="developers.html#BRANCHES"
>Section 3.3</A
>) or only one version
released at a time, my experience with several free software
projects and with the Debian project has taught me that use of
Linux's version numbering system is worth taking into
consideration. In Debian, <EM
>all</EM
> minor
versions are stable distributions (2.0, 2.1, etc). However,
many people assume that 2.1 is an unstable or development
version and continue to use an older version until they get so
frustrated with the lack of development progress that they
complain and figure the system out. If you never release an odd
minor version but only release even ones, nobody is hurt, and
less people are confused. It's an idea worth taking into
consideration.
</P
></DD
><DT
>Wine version numbering:</DT
><DD
><P
>Because of the unusual nature of wine's development where
the not-emulator is constantly improving but not working toward
any immediately achievable goal, wine is released every three
weeks. Wine does this by labeling their releases in <SPAN
CLASS="QUOTE"
>"Year
Month Day"</SPAN
> format where each release might be labeled
<SPAN
CLASS="QUOTE"
>"wine-XXXXXXXX"</SPAN
> where the version from January 04,
2000 would be <SPAN
CLASS="QUOTE"
>"wine-20000104"</SPAN
>. For certain
projects, <SPAN
CLASS="QUOTE"
>"Year Month Day"</SPAN
> format can make a lot of
sense.
</P
></DD
><DT
>Mozilla milestones:</DT
><DD
><P
>When one considers Netscape 6 and vendor versions, the
mozilla's project development structure is one of the most
complex free software models available. The project's version
numbering has reflected the unique situation in which it is
developed.
</P
><P
> Mozilla's version numbering structure has historically been
made up of milestones. From the beginning of the mozilla
project, the goals of the project in the order and degree to
which they were to be achieved were charted out on a series of
<A
HREF="http://www.mozilla.org/roadmap.html"
TARGET="_top"
>road
maps</A
>. Major points and achievements along these
road-maps were marked as milestones. Therefore, although
Mozilla was built and distributed nightly as <SPAN
CLASS="QUOTE"
>"nightly
builds,"</SPAN
> on a day when the goals of a milestone on the
road-map had been reached, that particular build was marked as
a <SPAN
CLASS="QUOTE"
>"milestone release."</SPAN
>
</P
><P
> While I haven't seen this method employed in any other projects
to date, I like the idea and think that it might have value in
any testing or development branch of a large application under
heavy development.
</P
></DD
></DL
></DIV
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="DOCUMENTATION">2.5. Documentation</H2
><P
> A huge number of otherwise fantastic free software applications
have withered and died because their author was the only person
who knew how to use them fully. Even if your program is written
primarily for a techno-savvy group of users, documentation is
helpful and even necessary for the survival of your project. You
will learn later in <A
HREF="users.html#RELEASING"
>Section 4.3</A
> that you should
always release something that is usable. <EM
>A piece of
software without documentation is not usable.</EM
>
</P
><P
> There are lots of different people you should document for and
there are lots of ways to document your project. <EM
>The
importance of documentation in source code to help facilitate
development by a large community is vital</EM
> but it falls
outside the scope of this HOWTO. This being the case, this section
deals with useful tactics for user-directed documentation.
</P
><P
> A combination of tradition and necessity has resulted in a
semi-regular system of documentation in most free software
projects that is worth following. Both users and developers expect
to be able to get documentation in several ways and it's essential
that you provide the information they are seeking in a form they
can read if your project is ever going to get off the
ground. People have come to expect:
</P
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN340">2.5.1. Man pages</H3
><P
>Your users will want to be able to type <SPAN
CLASS="QUOTE"
>"man
yourprojectname"</SPAN
> end up with a nicely formatted man page
highlighting the basic use of your application. Make sure that
before you release your program, you've planned for this.
</P
><P
> Man pages are not difficult to write. There is excellent
documentation on the man page writing process available through
the <SPAN
CLASS="QUOTE"
>"The Linux Man-Page-HOWTO"</SPAN
> which is available
through the Linux Documentation project <SPAN
CLASS="ACRONYM"
>(LDP)</SPAN
>
and is written by Jens Schweikhardt. It is available <A
HREF="http://www.schweikhardt.net/man_page_howto.html"
TARGET="_top"
>from
Schweikhardt's site</A
> or <A
HREF="http://www.linuxdoc.org/HOWTO/mini/Man-Page.html"
TARGET="_top"
>from the
<SPAN
CLASS="ACRONYM"
>LDP</SPAN
></A
>.
</P
><P
> It is also possible to write man pages using DocBook
SGML. Because man pages are so simple and the DocBook method
relatively new, I have not been able to follow this up but would
love help from anyone who can give me more information on how
exactly how this is done.
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN351">2.5.2. Command line accessible documentation</H3
><P
> Most users will expect some basic amount of documentation to be
easily available from the command line. For few programs should
this type of documentation extend for more than one screen (24 or
25 lines) but it should cover the basic usage, a brief (one or
two sentence) description of the program, a list of the commands
with explanations, as well as all the major options (also with
explanations), plus a pointer to more in-depth documentation for
those who need it. The command line documentation for Debian's
apt-get serves as an excellent example and a useful model:
</P
><TABLE
BORDER="1"
BGCOLOR="#E0E0E0"
WIDTH="100%"
><TR
><TD
><FONT
COLOR="#000000"
><PRE
CLASS="SCREEN"
>apt 0.3.19 for i386 compiled on May 12 2000 21:17:27
Usage: apt-get [options] command
apt-get [options] install pkg1 [pkg2 ...]
apt-get is a simple command line interface for downloading and
installing packages. The most frequently used commands are update
and install.
Commands:
update - Retrieve new lists of packages
upgrade - Perform an upgrade
install - Install new packages (pkg is libc6 not libc6.deb)
remove - Remove packages
source - Download source archives
dist-upgrade - Distribution upgrade, see apt-get(8)
dselect-upgrade - Follow dselect selections
clean - Erase downloaded archive files
autoclean - Erase old downloaded archive files
check - Verify that there are no broken dependencies
Options:
-h This help text.
-q Loggable output - no progress indicator
-qq No output except for errors
-d Download only - do NOT install or unpack archives
-s No-act. Perform ordering simulation
-y Assume Yes to all queries and do not prompt
-f Attempt to continue if the integrity check fails
-m Attempt to continue if archives are unlocatable
-u Show a list of upgraded packages as well
-b Build the source package after fetching it
-c=? Read this configuration file
-o=? Set an arbitary configuration option, eg -o dir::cache=/tmp
See the apt-get(8), sources.list(5) and apt.conf(5) manual
pages for more information and options.
</PRE
></FONT
></TD
></TR
></TABLE
><P
> It has become a GNU convention to make this type of information
accessible with the <SPAN
CLASS="QUOTE"
>"-h"</SPAN
> and the
<SPAN
CLASS="QUOTE"
>"--help"</SPAN
> options. Most GNU/Linux users will expect
to be able to retrieve basic documentation these ways so if you
choose to use different methods, be prepared for the flames and
fallout that may result.
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN358">2.5.3. Files users will expect</H3
><P
> In addition to man pages and command-line help, there are certain
files where people will look for documentation, especially in any
package containing source code. In a source distribution, most of
these files can be stored in the root directory of the source
distribution or in a subdirectory of the root called
<SPAN
CLASS="QUOTE"
>"doc"</SPAN
> or <SPAN
CLASS="QUOTE"
>"Documentation."</SPAN
> Common files
in these places include:
</P
><P
> <P
></P
><DIV
CLASS="VARIABLELIST"
><DL
><DT
>README or Readme</DT
><DD
><P
>A document containing all the basic installation,
compilation, and even basic use instructions that make up the
bare minimum information needed to get the program up and
running. A README is not your chance to be verbose but should
be concise and effective. An ideal README is at least 30 lines
long and more no more than 250.</P
></DD
><DT
>INSTALL or Install</DT
><DD
><P
>The INSTALL file should be much shorter than the README
file and should quickly and concisely describe how to build
and install the program. Usually an INSTALL file simply
instructs the user to run <SPAN
CLASS="QUOTE"
>"./configure; make; make
install"</SPAN
> and touches on any unusual options or actions
that may be necessary. For most relatively standard install
procedures and for most programs, INSTALL files are as short
as possible and are rarely over 100 lines.</P
></DD
><DT
>CHANGELOG, Changelog, ChangeLog, or changelog</DT
><DD
><P
>A CHANGELOG is a simple file that every well-managed
free software project should include. A CHANGELOG is simple
the file that, as its name implies, logs or documents the
changes you make to your program. The most simple way to
maintain a CHANGELOG is to simply keep a file with the source
code for your program and add a section to the top of the
CHANGELOG with each release describing what has been changed,
fixed, or added to the program. It's a good idea to post the
CHANGELOG onto the website as well because it can help people
decide whether they want or need to upgrade to a newer version
or wait for a more significant improvement.</P
></DD
><DT
>NEWS</DT
><DD
><P
>A NEWS file and a ChangeLog are similar. Unlike a
CHANGELOG, a NEWS file is not typically updated with new
versions. Whenever new features are added, the developer
responsible will make a note in the NEWS file. NEWS files
should not have to be changed before a release (they should be
kept up to date all along) but it's usually a good idea to
check first anyway because often developers just forget to
keep them as current as they should.</P
></DD
><DT
><SPAN
CLASS="ACRONYM"
>FAQ</SPAN
></DT
><DD
><P
>For those of you that don't already know,
<SPAN
CLASS="ACRONYM"
>FAQ</SPAN
> stands for Frequently Asked Questions
and a FAQ is a collection of exactly that. FAQs are not
difficult to make. Simply make a policy that if you are asked
a question or see a question on a mailing list two or more
times, add the question (and its answer) to your FAQ. FAQs are
more optional than the files listed above but they can save
your time, increase usability, and decrease headaches on all
sides.</P
></DD
></DL
></DIV
>
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN388">2.5.4. Website</H3
><P
> It's only indirectly an issue of documentation but a good website
is quickly becoming an essential part of any free software
project. Your website should provide access to your documentation
(in <SPAN
CLASS="ACRONYM"
>HTML</SPAN
> if possible). It should also include
a section for news and events around your program and a section
that details the process of getting involved with development or
testing and make an open invitation. It should also supply links
to any mailing lists, similar websites, and provide a direct link
to all the available ways of downloading your software.
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN392">2.5.5. Other documentation hints</H3
><P
></P
><UL
><LI
><P
> All your documentation should be in plaintext, or, in cases
where it is on your website primarily, in HTML. Everyone can
cat a file, everyone has a pager, (almost) everyone can render
HTML. <EM
>You are welcome to distribute information in
PDF, PostScript, RTF, or any number of other widely used
formats but this information must also be available in
plaintext or HTML or people will be very angry at
you.</EM
> In my opinion, info falls into this category
as well. There is plenty of great GNU documentation that
people simply don't read because it only in info. And this
<EM
>does</EM
> make people angry. It's not a
question of superior formats; it is a question of
accessability and the status quo plays a huge role in this
determination.
</P
></LI
><LI
><P
> It doesn't hurt to distribute any documentation for your
program from your website (FAQs etc) with your program. Don't
hesitate to throw any of this in the program's tarball. If
people don't need it, they will delete it. I can repeat it over
and over: <EM
>Too much documentation is not a
sin.</EM
>
</P
></LI
><LI
><P
>Unless your software is particular to a non-English
language (a Japanese language editor for example), please
distribute it with English language documentation. If you don't
speak English or not not confident in your skills, ask a friend
for help. Like it or not, fair or unfair, <EM
>English is
the language of free software</EM
>. However, this does not
mean you should limit your documentation to only English. If you
speak another language, distribute translations of documentation
with your software if you have the time and energy to do
so. They will invariably be useful to someone.</P
></LI
><LI
><P
> Finally, <EM
>please spell-check your
documentation.</EM
> Misspellings in documentation are
bugs. I'm very guilty of committing this error and it's
extremely easy to do. If English is not your first language,
have a native speaker look over or edit your documentation or
web pages. Poor spelling or grammar goes a long way to making
your code look unprofessional. In code comments, this type of
thing is less important but in man pages and web pages these
mistakes are not acceptable.
</P
></LI
></UL
></DIV
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="PRESENTATION">2.6. Other Presentation Issues</H2
><P
> Many of the remaining issues surrounding the creation of a new
free software program fall under what most people describe as
common sense issues. Its often said that software engineering is
90 percent common sense combined with 10 percent specialized
knowledge. Still, they are worth noting briefly in hopes that they
may remind a developer of something they may have forgotten.
</P
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN411">2.6.1. Package File Names</H3
><P
> I agree with ESR when he says that: <SPAN
CLASS="QUOTE"
>" It's helpful to
everybody if your archive files all have GNU-like names --
all-lower-case alphanumeric stem prefix, followed by a dash,
followed by a version number, extension, and other
suffixes."</SPAN
> There is more info (including lots of examples
of what <EM
>not</EM
> to do in his <I
CLASS="CITETITLE"
>Software
Release Practices HOWTO</I
> which is included in this
HOWTO's bibliography and can be found through the LDP.
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN417">2.6.2. Package formats</H3
><P
> Package formats may differ depending on the system you are
developing for. For windows based software, Zip archives (.zip)
usually serve as the package format of choice. If you are
developing for GNU/Linux, *BSD, or any UN*X, make sure that your
source code is always available in tar'ed and gzip'ed format
(.tar.gz). UNIX compress (.Z) has gone out of style and
usefulness and faster computers have brought bzip2 (.bz2) into
the spot-light as a more effective compression medium. I now make
all my releases available in both gzip'ed and bzip2'ed tarballs.
</P
><P
> Binary packages should always be distribution specific. If you
can build binary packages against a current version of a major
distribution, you will only make your users happy. Try to foster
relationships with users or developers of large distributions to
develop a system for the consistent creation of binary
packages. It's often a good idea to provide RedHat
<SPAN
CLASS="ACRONYM"
>RPM</SPAN
>'s (.rpm), Debian deb's (.deb) and source
<SPAN
CLASS="ACRONYM"
>RPM</SPAN
>'s <SPAN
CLASS="ACRONYM"
>SRPM</SPAN
>'s if
possible. Remember: <EM
>While these binaries packages are
nice, getting the source packaged and released should always be
your priority. Your users or fellow developers can and will do
the the binary packages for you.</EM
>
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN425">2.6.3. Version control systems</H3
><P
> A version control system can make a lot of these problems of
packaging (and a lot of other problems mentioned in this HOWTO)
less problematic. If you are using *NIX, CVS is your best bet. I
recommend Karl Fogel's book on the subject (and the <A
HREF="http://cvsbook.red-bean.com/"
TARGET="_top"
>posted HTML version</A
>)
wholeheartedly.
</P
><P
> CVS or not, you should probably invest some time into learning
about a version control system because it provides an automated
way of solving many of the problems described by this HOWTO. I
am not aware of any free version control systems for Windows or
Mac OS but I know that CVS clients exist for both
platforms. Websites like <A
HREF="http://sourceforge.net"
TARGET="_top"
>SourceForge</A
> do a great job
as well with a nice, easy-to-use web interface to CVS.
</P
><P
> I'd love to devote more space in this HOWTO to CVS because I love
it (I even use CVS to keep versions straight on this HOWTO!) but
I think it falls outside the scope of this document and already
has its own HOWTOs. Most notably is the <I
CLASS="CITETITLE"
>CVS Best
Practices HOWTO</I
><A
HREF="b811.html#CVSBESTPRACTICES"
>[CVSBESTPRACTICES]</A
>
which I've included in the attached bibliography.
</P
></DIV
><DIV
CLASS="SECT3"
><H3
CLASS="SECT3"
><A
NAME="AEN434">2.6.4. Useful tidbits and presentation hints</H3
><P
> Other useful hints include:
</P
><P
> <P
></P
><UL
><LI
><P
> <EM
>Make sure that your program can always be found in a
single location.</EM
> Often this means that you have a
single directory accessible via <SPAN
CLASS="ACRONYM"
>FTP</SPAN
> or the
web where the newest version can be quickly recognized. One
effective technique is a provide a symlink called
<SPAN
CLASS="QUOTE"
>"yourprojectname-latest"</SPAN
> that is always pointing
to the most recent released or development version of your
free software application. Keep in mind that this location
will receive many requests for downloads around releases so
make sure that the server you choose has adequate bandwidth.
</P
></LI
><LI
><P
> <EM
>Make sure that there is a consistent email address
for bug reports.</EM
> It's usually a good idea to make
this something that is NOT your primary email address like
yourprojectname@host or yourprojectname-bugs@host. This way,
if you ever decide to hand over maintainership or if your
email address changes, you simply need to change where this
email address forwards. It also will allow for more than one
person to deal with the influx of mail that is created if your
project becomes as huge as you hope it will.
</P
></LI
></UL
>
</P
></DIV
></DIV
></DIV
><DIV
CLASS="NAVFOOTER"
><HR
ALIGN="LEFT"
WIDTH="100%"><TABLE
SUMMARY="Footer navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
><A
HREF="intro.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="index.html"
ACCESSKEY="H"
>Home</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
><A
HREF="developers.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
>Introduction</TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
>&nbsp;</TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
>Maintaining a Project: Interacting with Developers</TD
></TR
></TABLE
></DIV
></BODY
></HTML
>
Reference in New Issue View Git Blame Copy Permalink