Linux WWW-HOWTO
by Peter Dreuw, pdreuw@wing.gun.de
v0.7.6, 6 October 1996
This document contains information about setting up WWW services under
Linux (both server and client) and how to maintain them. It tries not
to be a in detail manual but an overview and a good pointer to further
information.
1. Introduction
Many people are steping into Linux 'cause they are looking for a
really good internet capable platform. Others use Linux for the fun
installing a free OS on their system. Some of those want to get in
touch with the internet, of course. Furthermore, there are institutes,
universities and other mostly not-for-profit organisations which want
to or need to set up internet sites on small expenses. This is, where
the WWW HowTo comes in. This document tries to explain how to set up
clients and servers for the (in my mind) largest online part of the
net - The World Wide Web.
1.1. Copyright
This document is Copyright (c) 1996 by Peter Dreuw. Please copy and
distribute it widely, but do not modify the text or omit my name.
If you sell this HOWTO on a CD, in a book or on another media, I would
really like to have a copy for reference.
Trademarks are owned by there respective owners.
1.2. Disclaimer
This document is meant as an introduction into WWW techniques used or
usable on Linux. I an not an WWW nor security expert ! I AM NOT
RESPONSIBLE FOR ANY DAMAGES INCURRED DUE TO ACTIONS TAKEN BASED ON THE
INFORMATION INCLUDED IN THIS DOCUMENT.
1.3. Feedback
Any feedback is really welcome. Just mail to pdreuw@wing.gun.de.
1.4. New versions of this Document
New versions of this document can be retrieved via anonymous FTP from
sunsite.unc.edu under /pub/Linux/docs/HOWTO and almost any friendly
Linux ftp mirror site.
Furthermore, you can download it via
<http://ourworld.compuserve.com/homepages/dreuw/lxwwwh2.tgz> as
gzipped tar archive containing a sgml, text, latex and ps version.
The html version is directly available under
<http://ourworld.compuserve.com/homepages/dreuw/lxwwwh2.htm>
2. Setting up WWW client software
The following chapter is dedicated to the web users. Some hacks and
tricks setting up current versions of common web browsers. Please feel
free to contact me, if your favorite web browser is not mentioned
here. (As this is a really early version of the WWW-HOWTO, most of
them are likely not to be listed...)
Personally, I prefer the Emacs - W3 browser and Lynx as they have some
speed advantages and there is no need to retrieve the complete
graphics through my slow speeded dial up line ;)
2.1. Overview
Lynx is the smallest Web browser I know and use - but ist has many
special features, so don't skip this chapter.
Emacs - well there is nothing to say about the Emacs W3 browser, its
just Emacs, like the Emacs news reader, the Emacs mail reader etc. pp.
Netscape Navigator is the only browser mentioned here, which is
capable of this new funny things like JavaScript and these nice
<APPLET> tag feature needed tu run Java. Please report if there is any
other web browser which can do the one or other. I'd really like to
know.
There are rumors, that Microsoft is going to port the Internet
Explorer to varios Unix platforms - maybe including Linux. If you DO
know something more reliable, please drop me a mail.
2.2. Lynx
The smallest (?, hm, something around 650 K executable) and maybe
fastest Web browser available. It does not eat up much bandwidth nor
system resources as it only deals with text displays like any console,
terminal or xterm. You don't need any X Window system nor additional
megabytes of system memory running this little browser.
Furthermore, the source code is available, too.
2.2.1. Where to get
The latest version is 2.5 and can be retrieved from
<http://www.wfbr.edu/dir/lynx> or from almost any friendly Linux ftp
server like ftp://sunsite.unc.edu under
/pub/Linux/system/Network/info-systems/www/ or mirror site.
Or, take a look at the Lynx enhanced pages
<http://www.nyu.edu/pages/wsn/subir/lynx.html> for information on
using Lynx.
2.2.2. How to install
Just retrieve the archive, unpack it, read the README and follow the
steps told in the INSTALLATION file.
If you don't want a source distribution, you'd maybe retrieve a binary
distribution for the Linux on Intel based systems available on
sunsite.
Lynx compiles and runs on my system without any problems on both Linux
1.2.13 and 2.0.x.
2.2.3. Special features
Well, there are. For a complete description, just read the manuals and
doc files that come with Lynx.
To get a nice glimpse, just type in
lynx --help
and be impressed.
In my humble opinion, the most special feature of Lynx against all
other web browsers is the capability for batch mode retrival. One can
write a shell script which retrieves a document, file or anything like
that via http, ftp, gopher, WAIS, NNTP or file:// - url's and save it
to disk. Furthermore, one can fill in data into HTML forms in batch
mode by simply redirecting the standard input and using the -post_data
option.
2.3. Emacs-W3
There is one sad thing about the Emacs W3 browser ;) If you got GNU
Emacs or XEmacs running, you probably got the W3 browser running to.
Not much work in this HOWTO. If you feel, that there should be more
information about this, please let me know.
The Emacs W3 mode is a nearly fully featured web browser system
written in the Emacs Lisp system. It mostly deals with text, but can
display graphics, too - at least - if you run the emacs under the X
Window system.
The most recent GNU emacs package is available under
<ftp://prep.mit.ai.edu>, the most recent XEmacs could be retrieved
from <ftp://ftp.xemacs.org>.
2.4. Netscape Navigator Gold 3.0
Yeah, you made it. The Queen of WWW browsers. Something almost like
Emacs is in the world of text editors. Netscape Navigator can do
nearly everything (except cooking coffee... but maybe java will
do...). But on the other hand, the most memory hungry and resource
eating pice of web browser, news reader, mail reader (pop3), mail &
news editor I've ever seen.
My latest version of the Netscape Navigator Gold (export version) is
from 28-Aug-1996 and (c) 1995, 1996 Netscape Communications Corp.
(As I live in Europe, I can only get the export version...)
2.4.1. Where to get
The first place to get the Netscape Navigator for Linux as binary
distribution is on <ftp://ftp.netscape.com>. The second - as these
servers are heavily loaded - may be any friendly netscape mirror site.
You might as well ask archie about this. Maybe, you'll be happy and
find it on a cd rom - this will save some bandwidth as the archive is
quite large ( 2.5 MB).
2.4.2. Unpacking & Installing
Unpack the archive und read the README file ! There is really nothing
strange about this, you know.
2.4.3. Java applets with the navigator
There are some reports telling that there are problems running java
applets with the Netscape Navigator Gold 3.0 even if java is activated
in the otions dialog. The archive known to me contained a file
java_30 which must be renamed to java_30.zip. After this, any java
applet should work fine within the netscape environment.
If you continue to have problems using java applets like Netscape
Navigator hangs or just terminates after downloading a java applet,
take a look at your libc version. Just do a
ldconfig -v | less
(maybe, you have to be root doing so...) and watch out for an entry
libc.so.5 => libc.so.5.xx.yy
where your libc version is 5.xx.yy. If your libc isn't 5.2.18, this
may be the problem. There are many reports for Linux 1.2.13 systems,
that they should upgrade to libc 5.2.18 when the need to run Netscape
Navigator in general. Additionally, it may be a good idea to downgrade
your libc from 5.3.xx to the 5.2.18 if you run Netscape Navigator and
a Linux 2.0.x kernel. (In fact, the libc 5.3.xx series is for beta
testing purposes, so you should know what you're doing.) Some of the
5.3.xx series break Netscape Navigator and the Java classes code.
For more information on Java on Linux or Java programming, please read
the JAVA-HOWTO or visit <http://www.sun.com>.
3. Setting up WWW server systems
This section contains information on different http server software
packages and additional server side tools like script languages for
CGI programs etc.
For a technical description on the http mechanism, take a look at the
RFC documents menitoned in the chapter "For further reading" of this
HOWTO.
3.1. cern httpd
As the cern original httpd server is reported to have some ugly bugs,
to be quite slow and resource hungry, it is not described in this
HOWTO by now. If you volunteer to admit some facts or chapters, please
send them to me, I'll add them to this doc.
3.2. apache
-To be written - sorry Features, Overview, Advantages
3.2.1. Where to get
3.2.2. Installing
3.2.3. Configuring
3.2.4. Special Features
Apache httpd has got some special features in the actual version.
3.2.4.1. Host multicasting
BlaBla??? how to setup ....
3.2.4.2. Module system
how to include other modules ... where to get infos about module
programming ...
3.3. CGI scripts systems
- to be written - sorry - CGI (common gateway interface)
3.3.1. How does CGI work in principle ?
- to be written - sorry - calling structure, http structure, program
parameter format (slightly touched), things to keep in mind
3.3.2. Perl
- to be written - sorry - something easy in perl (sample script)
3.3.3. PHP/FI
- to be written - sorry - something easy in PHP/FI (sample script)
3.3.4. W3-mSQL
- to be written - sorry - something even more easy (sample script)
hint about setting up !!!
3.3.5. some useful scripts
- to be written - sorry -FaxInbound to nice Table including php/fi
script and shell script
4. Maintaining a WWW site or some Web Pages
If you have to maintain a web site or if you maintain at least a web
page, you have to think about your offer to the network and you have
to spend some thoughts about approaching the reader / user of your web
pages.
4.1. The mainstream: HTML technical
Well, I'm not gonna tell you, how HTML is encoded an how you have to
design your pages. I'll just give you some pointers where you can find
more advanced information.
You should take a look at <http://www.w3.org/> for the latest HTML
language specification.
Take a look at the list at the end of this article, you'll find more
hints, where to read on.
4.2. Some thoughts about bandwidth
Many users connect to the internet via slow speed modem lines. A
speed range from 14,400 bps to 28,800 bps is state-of-art for "private
sites". In europe, there are ISDN systems growing, but a speed of
64,000 bps isn't that more fast in comparison to - let's keep it
simple - 10,000,000 bps ethernet. And 10 Mbps ethernet isn't really a
high speed LAN connection nowadays.
As you realize that many users don't have this fast access to the net,
you should keep in mind to put up the relation between information and
bytes. Optimize it at 1:1 - if you can. You may use graphics in your
web pages following the multi media trend, but always remember the
goals of your page and of the graphic you're going to put in. If most
of your users are connected via a small modem line and the graphic
severes only for estethic reasons or some eye-catching effects, you'd
better bann it from your pages, or -at least- rerender it to the
smallest possible file size and use best compression. Your users will
like it.
Always remember, nobody really likes an eye-catcher, that comes up
about 3-5 minutes after the text message.
4.3. Some thoughts about server load
On a web server, there is normally at least one server task running.
If this task reads a request from a http client, it duplicates itself
(on Linux it's called forking) and the new copy serves the request,
while the original keeps listening for new requests. After finishing
the request, the copy terminates. (In fact, some servers - like the
apache - always keep up a default of five ready waiting server copies
for requests parallel to the master incarnation for speed reasons.)
Some web browsers like the Netscape Navigator series do many requests
parallel on the same server, which increases the server load spend on
the same user. These browsers e.g. retrieve the HTML page and parse
them while retrieving and issue new requests for other information
like the embedded graphics, applet files, sound files or any other
additional mime-encoded data. In opposition, 'simple' browsers request
and retrieve one file after another, which keeps the server per user
load relation as low as possible.
Many users prefer browsers that use the multi request technique like
the Netscape Navigator, because they bring up a more complete overview
on the requested page before the single request browser does.
This is in my opinion because many page designers do stick on
embedding the information into the graphics, denying the text-only
browsers.
So, we - as server maintainers - got the problem, that most of the
users cast multiple requests on out server within the same page
retrival. We can limit this by limiting the server software not to
serve more requests than "x" from the same requesting system at the
same time. But how to get this "x" ? It's not easy to calculate and a
lot of personal expirience on your site is necessary to depict it. But
I'll give you some hints. We have to take our connection bandwidth
into account, our server memory size, some feeling about our servers
cpu/disk performance and ... well, that's enough for the first
glimpse. You should take a look at the memory usage a single server
task has. Then think, how many of them could kept in memory at all.
Think, how many per cents of your web pages could remain in your
servers disk cache. Optimize the count of web server tasks against the
disk cache size and you're really near to your personal "x".
Furthermore, you can put in other jobs the server got. E.g. if your
system also serves for ftp, you might limit the maximum possible
connections to keep up some minimum room for the ftp server task. If
your web server also does some database services, you'd better keep up
some cpu cycles and also shrink your "x". Play somewhat around with
these values and test them. And (!) read the following chapter about
CGI scripting, which also takes server performance and - depending on
the CGI jobs - amount of memory.
4.4. CGI vs. Applet / Client side script
- to be written - sorry - overview ond advantage/disadvantage and
hints when to use which.
4.5. Style ideas
Uh, a really difficult theme for beeing on a short sentence. I don't
try to mix up your genious design ideas. Nor I'm gonna put you into my
personal design strategies. I'd just like to add one or two statements
to the above ideas on server load and bandwidth.
Numerous research on human behavior on user interfaces and on-screen
presentation have brought out interesting results. There are some
simple facts one should keep in mind designing WWW pages.
� Keep text in short blocks. This HOWTO is ugly to read on screen,
but nice to read in paper print. (Try it yourself!) Human beeings
often have difficulties to read lengthy text printed on screen.
They loose their point in the sentence; their concentration
suffers.
� Don't mix up graphics and text blocks. This is a good-looking but
ugly-to-read feature. You can spread Headlines, eyecatchers but,
please, don't mix up block text with graphics. Behaviorists found
out, that human are much more attracted by graphis on screen than
by text. People find it more easy to realize a graphic on screen
than on paper, in opposition to text which is more easy to "see and
decode" on paper than screen.
Did you know this ? If you'd like to get more information on that,
search for GUI style guides and ergonomy research results done by many
universities and software companies (including MS).
4.6. HTML editors under Linux
Hm, there are some. In fact, there are reported to be many. But as I
already shot my shoot, I didn't test them all. But I am really
curiosly looking forward to read the reports you're gonna mail.
4.6.1. vi, vim
vi and vim are perfectly usable for writing HTML code... (don't flame
me on that) because HTML code only uses ASCII text chars. I don't want
to give stuff for another editor war. Those who know vi/vim and use it
daily can use it for HTML code either. You can make vi/vim help you
developing HTML code by doing some macros for vi/vim. But as this is
no VI-HOWTO, I'll leave this fact alone here. Just take it, that it is
possible to use vi/vim for HTML editing (at least for some short
changes). If you already know how to program vi/vim, you'll certainly
know how to abstract for HTML either. If you don't do so, well, don't
care.
4.6.2. emacs & XEmacs
- to be written - sorry -
4.6.3. asWedit
- to be written - sorry -
4.6.4. other pointers
Ah, there was some reference for a package named phoenix, based on
tkWWW, but I was not able to get them running on my system. I think,
it was a problem with my tcl/tk versions but you'll never know. I
didn't spend much time around with them, so, maybe they'll run on your
system both. Just go'n ask archie. Maybe, you can drop me a mail, if
you are sucessful.
If you miss your faivorite HTML editor here, just write a mail to me.
Maybe, I'll add some pointers to web pages about HTML editors for
Linux to. Just send me some nice URL's.
4.7. Graphics
Thoughts, Ideas, Hints ? Well, you may read the comp.graphics
newsgroup. And, you can visit <http://www.w3.org/pub/WWW/Graphics/>.
4.7.1. Format gif
GIF (Graphics Interchange Format) was introduced 1987 by Compuserve,
Inc. an revised 1989. It uses a LZ algorythm, which underlies U.S.
copyright or patent law. So there might exist some legal problems
using this graphics format in the internet - despite the fact that
nearly anybody does.
Gif is a good format for small pictures with simple structured
graphics like computer graphics or banners.
Gif has some advantages as it is one of the (if not the) widest spread
graphic formats in online systems:
� offers a good compression
� compresses without information loss
� has a interlace capability, i.e. pictures could be viewed in full
size (with less resolution) before they're retrieved completely.
� can hold more than one picture within one file
� can hold a small animation in one file
� nearly any graphical web browser supports gif
� can hold a transparent color
� fast decompression system
The disadvantages are:
� only 256 color pictures possible
� license and copyright problems (?)
� not ideal file size
4.7.2. Format jpeg
The Joint Graphic Experts Group (JPEG) did the design for the
jpeg/jpg/jiff graphic format. This format is based on a discrete
cosinus transformation (DCT) and a Huffmann encode compression. JPEG
works with an significant information loss, which can make your
pictures somewhath less colorous or less sharp. Typical compression
factor is 1:5 ranging to 1:50. (Above 1:10 anybody is able to see the
artefacts risen through the compression/decompression cyle.)
JPEG is a good format for photographies, large graphics and really
complex pictures.
The advantages are:
� strong compression, small files and therefor fast download...
� any graphical browser knows about jpeg
The disadvantages are:
� slow compression/decompression
� possible information loss
4.7.3. Format png
Portable Network Graphics (PNG) - the new format on the net. PNG is
favorised by the W3 consortium. For some more special information
visit <http://www.w3.org/pub/WWW/TR/WD-png.html> and
<http://www.w3.org/pub/WWW/Graphics/PNG/Overview.html>. Here you'll
find a technical specification, some programmers information etc. PNG
is a ideal format replacing GIF. The PNG homepage is on
<http://quest.jpl.nasa.gov/PNG/>. For the users, PNG will have some
advantages and some disadvantages. Here they are:
For the advantages:
� can replace the license loaded GIF - PNG has no license problems
� 256 palette system as well as grayscale and true color capability
including a transparency element
� complex interlace mode where not only different lines are sequenced
but a two dimensional serialize system retrieves the picture
resulting the user to realise the picture content more early.
� fast decompression algorythm is possible
� public available description - license free
� public available sample code - license free
� extensible design
For the disadvantages:
� not widely spread (Netscape does not support it by now, some
plugins do)
� not so strong compressing pictures
� no final specification ready, in working draft state.
PNG is currently supported on Linux through the following programs:
ImageMagick (Version >=3.7), GhostScript 4.0, Gimp, PovRay 3.0, the
netpbm package. For xv 3.10a there exists an inofficial patch.
4.7.4. Converters
- to be written - sorry - netpbm, xv, ghostscript, gimp, ImageMagick,
CorelDraw auf Wine :-)))
4.8. Specials
There are now many specials beyond the HTML'n'Image range. There are
Applets written in Java and JavaScript pages and many things beyond.
4.8.1. Java
There is nothing to add about Java in general, just read the java
section in the Netscape Navigator chapter of this HOWTO and the
overview on Java Applett vs. CGI script in this HOWTO. Then, you can
also read the really good and compact Linux JAVA HOWTO. For
programming Java, please refer really good books on that.
4.8.2. ActiveX
ActiveX is at the time of writing still a Microsoft child. Microsoft
claimed, that they would release it to the public domain or at least
to release it to a ActiveX consortium.
ActiveX has nothing to do with the X Window system nor with XFree.
It is derived from the Microsoft and IBM OLE system. After releasing
the specs, there should be a Unix port. But, we have to wait till
then. Nothing for Linux, yet.
5. FAQ
There aren't any frequent asked questions - yet...
6. For further reading
� RFC1866 written by T. Berners-Lee and D. Connolly, "Hypertext
Markup Language - 2.0", 11/03/1995
� RFC1867 writtenm by E. Nebel and L. Masinter, "Form-based File
Upload in HTML", 11/07/1995
� RFC1942 written by D. Raggett, "HTML Tables", 05/15/1996
� RFC1945 by T. Berners-Lee, R. Fielding, H. Nielsen, "Hypertext
Transfer Protocol -- HTTP/1.0", 05/17/1996.
� RFC1630 by T. Berners-Lee, "Universal Resource Identifiers in WWW:
A Unifying Syntax for the Expression of Names and Addresses of
Objects on the Network as used in the World-Wide Web", 06/09/1994
� RFC1959 by T. Howes, M. Smith, "An LDAP URL Format", 06/19/1996
7. Thanks
Special thanks to Greg Hankins gregh@cc.gatech.edu for encuraging me
to write this work and the fun I had doing it.
I'd also like to thank Chris Hendricks, Fido: 2:2433/443@fidonet.org
for his engagement in Linux and my personal race to keep at least one
nose ahead :-)