aboutsummaryrefslogblamecommitdiffstats
path: root/doc/STORY.html
blob: 8e89f8c64df33900317e99b45056005a2358da33 (plain) (tree)















































































































































































































                                                                                                                                     
<html>
<head>
<title>History of w3m</title>
</head>
<body>
<h1>History of w3m</h1>
<div align=right>
1999/2/18<br>
1999/3/8 revised<br>
1999/6/11 translated into English<br>
Akinori Ito<br>
aito@fw.ipsj.or.jp
</div>
<h2>Introduction</h2>
W3m is a text-based pager and WWW browser.
It is similar application to the famous text-based
browser <a href="http://www.lynx.browser.org/">Lynx</a>.
However, w3m has several advantages against Lynx. For example,
<UL>
<LI>W3m can render tables.
<LI>W3m can render frame (by converting frame into table).
<LI>As w3m is a pager, it can read document from standard input.
(I heard Lynx also can display standard-input-given document, like this:
<pre>
   lynx /dev/fd/0 &gt; file
</pre>
Hmm, it works on Linux. )
<LI>W3m is small. Its stripped binary for Sparc (compiled with
gcc -O2, version beta-990217) is only 260kbyte, while binary size
of Lynx is beyond 1.8Mbyte. (Actually, lynx it 800K on my i386 system, w3m is 200K + libgc.)
</UL>
It is true that Lynx is an excellent browser, who have many
features w3m doesn't have. For example,
<UL>
<LI>Lynx can handle cookies.
<LI>Lynx has many options.
<LI>Lynx is multilingual. (W3m is Japanese-English bilingual)
</UL>
etc. It is also a great advantage that Lynx has a lot of
documentation.
<P>
<b>I don't intend w3m to be a substitute of any other browsers,
including Netscape and Lynx.</b> Why did I wrote w3m?
Because I felt inconvenient with conventional browsers 
to `take a look' at web pages.
I am browsing web pages in LAN environment. When I want to take
a glance at a web page, I don't want to wait to start up Netscape.
Lynx also takes a few seconds to start up (you can get lynx startup time to almost zero when you rm /etc/mailcap). On the other hand,
w3m starts immediately with little load to the host machine.
After looking at the information using w3m, I use other browser
if I want to read the the page in detail. As for me, however,
w3m is enough to read most of web pages.

<h2>The birth of w3m</h2>
<P>
w3m was derived from a pager named `fm'. Fm was written before
1991 (I don't remember the exact date) when WWW was not popular.
At that time, the word `browser' meant a file browser like
`more' or `less'.
<P>
I wrote fm to debug a program for my research. To trace the status
of the program, it dumped megabytes of values of variables into a file,
and I debugged it by checking the dumped file. The program dumped
information at a certain time in one line, which made the dumped line
several hundred characters long. When I looked the file using `more' or
`less', one line was folded into several lines and it was very hard
to read it. Therefore, I wrote fm, which didn't fold a line. Fm displayed
one logical line as one physical line. When seeing the hidden
part of a line, fm shifted entire screen. As I used 80x24 terminal at that
time, fm was very useful for the debugging.
<P>
Several years later, I got to know WWW and began to use it.
I used XMosaic and Chimera. I liked Chimera because it was light.
As I was interested in the mechanism of WWW, I learned HTML and
HTTP, and I felt it simpler than I expected. The earlier version
of HTTP was very similar to Gopher protocol. HTML 2.0 was
simple enough to render. All I have to do seemed to be line folding
and itemized display. Then I made a little modification to fm
and made a web browser. It was the first version of w3m.
The name `w3m' was an abbreviation of Japanese phrase `WWW wo miru',
which means `see WWW'. It was an inheritance from `fm', which
was an abbreviation of `File wo miru'. The first version of w3m
was released at the beginning of 1995.

<h2>Death and rebirth of w3m</h2>
<p>
I had used w3m as a pager to read files, E-mails and online manuals. 
It was a substitute of less. Sometimes I used w3m as a web browser,
but there were many pages w3m couldn't display correctly, most of
which used table for page layout. Once I tried to implement table
renderer, but I gave up because it seemed to be too difficult for me.
<P>
It was 1998 when I tried to modify w3m again. There were two reasons.
The first is that I had some time to do it. I stayed Boston University
as a visiting researcher at that time. The second reason is that
I wanted to use table in my personal web page.  I had written research
log using HTML, and I wanted to write a table in it. At first I used 
&lt;pre&gt;..&lt;/pre&gt; to describe table, but it was not cool at all.
One day I used &lt;table&gt; tag, which made me to use Netscape to
read the research log. Then I decided to implement a table renderer
into w3m.
<P>
I didn't intend to write a perfect table renderer because tables
I used was not very complicated. However, incomplete table rendering
made the display of table-layout pages horrible. I realized that
it required almost-perfect table renderer 
to do well both in `rendering (real) table' and `fine display of
table-layout page.' It was a thorn path.
<P>
After taking several months, I finished `fair' table renderer.
Then I implemented form into w3m. Finally, w3m was reborn as a
practical web browser.

<h2>Table rendering algorithm in w3m</h2>

HTML table rendering is difficult. Tabular environment
of LaTeX is not very difficult, which makes the width of a column
either a specified value or the maximum width to put items into it.
On the other hand, HTML table renderer has to decide
the width of a column so that the entire table can fit into the
display appropriately, and fold the contents of the table according
to the column width. Inappropriate column width decision makes
the table ugly. Moreover, table can be nested, which makes the algorithm
more complicated.

<OL>
<LI>First, calculate the maximum and minimum width of each column.
The maximum width is the width required to display the column
without folding the contents. Generally, it is the length of 
paragraph delimited by &lt;BR&gt; or &lt;P&gt;.
The minimum width is the lower limit to display the contents.
If the column contains the word `internationalization', the minimum
width will be 20. If the column contains 
&lt;pre&gt;..&lt;/pre&gt;, the maximum width of the preformatted
text will be the minimum width of the column.

<LI>If the width of the column is specified by WIDTH attribute,
fix the column width using that value. If the specified width is
smaller than the minimum width of the column, fix the column width
to the minimum width.

<LI>Calculate the sum of the maximum width (or fixed width) of
each column and check if the sum exceeds the screen width.
If it is smaller than screen width, these values are used for
width of each column.

<LI>If the sum is larger than the screen width, determine the widths
of each column according to the following steps.
<OL>
<LI>Let W be the screen width subtracted by the sum of widths of 
fixed-width columns.
<LI>Distribute W into the columns whose width are not decided,
in proportion to the logarithm of the maximum width of each column.
<li>If the distributed width of a column is smaller than the minimum width,
then fix the width of the column to the minimum width, and 
do the distribution again.
</OL>
</OL>

In this process, distributed width is proportion to logarithm of
maximum width, but I am not sure that this heuristic is the best.
It can be, for example, square root of the maximum width.
<P>
The algorithm above assumes that the screen width is known.
But it is not true for nested table. According the algorithm above,
the column width of the outer table have to be known to render
the inner table, while the total width of the inner table have to
be known to determine the column width of the outer table.
If WIDTH attribute exists there are no problems. Otherwise, w3m
assumes that the inner table is 0.8 times as wide as the outer
table. It works fine, but if there are two tables side by side in an outer
table, the width of the outer table always exceeds the screen width.
To render this kind of table correctly, one have to render the table once,
check the width of outmost table, and then render the entire table again.
Netscape might employ this kind of algorithm.

<h2>Libraries</h2>

w3m uses
<a href="http://reality.sgi.com/boehm/gc.html">Boehm GC</a>
library. This library was written by H. Boehm and A. Demers.
I could distribute w3m without this library because one can
get the library separately, but I decided to contain it in the
w3m distribution for the convenience of an installer.
W3m doesn't use libwww.
<P>
Boehm GC is a garbage collector for C and C++. I began to use this
library when I implemented table, and it was great. I couldn't
implement table and form without this library. 
<P>
Older version than beta-990304 used 
<a href="http://home.cern.ch/~orel/libftp/libftp/libftp.html">LIBFTP</a>
because I felt tired of writing codes to handle FTP protocol.
But I rewrote the FTP code by myself to make w3m completely free.
It made w3m slightly smaller.
<P>
By the way, w3m doesn't use UNIX standard regexp library and curses library.
It is because I want to use Japanese. When I wrote fm, there were
no free regexp/curses libraries that can treat Japanese. Now both libraries
are available and they looks faster than w3m code.

<h2>Future work</h2>

...Nothing. As w3m's virtues are its small size and rendering speed,
adding more features might lose these advantages. On the other hand,
w3m is still known to have many bugs, and I will continue fixing them.

</body>
</html>