Practical Advice on Installing Papers on the Web
The great temptation, once you've finished the camera-ready copy of your conference or journal paper, is to move the postscript file to a public place and publish its URL. The result is often a paper with no page numbers, no indication of where it came from, and no way to update it or associate additional information with it.

To avoid these embarassments, I recommend the following steps.

## Restore Page Numbers

Most camera-ready material has no page numbers (these are added when the complete volume is assembled). Before publishing your paper on the web, turn page numbering on--there's nothing more annoying or amateurish-looking than a paper with no page numbers. In LaTeX, commenting out any \pagestyle{empty} and \thispagestyle{empty} commands will usually do the trick. If not, add \pagestyle{plain} to the prelude and \thispagestyle{plain} after the \maketitle. If those don't work, you'll have to study the actual style files and options that you're using and maybe enlist the help of a LaTeX expert (or read the manual).

## Add Bibliographic Data

It's a great paper! You want people to cite it! So why not help them to do so?

A paper that provides no information but its title and authorship is going to be cited (if at all) as "manuscript" or "unpublished notes" or "technical report" or something equally wounding. If it's a PODC paper, make sure the paper says so. There are various ways of adding bibliographic data to a paper, requiring varying degrees of LaTeX sophistication (I prefer one using fancyheadings.sty), but the bare-bones simplest is to add

\\ \small To be presented at Some Conference, Some City and Country, Date

to the title. If that extra line causes the pagination to mess up, stick a \vspace*{-1ex} or similar in like this
\title{\vspace*{-1ex}Original Title\thanks{sponsorship}
\\ \small To be presented at Some Conference, Some City and Country, Date}

Don't forget to update everything once the paper is published and you know the actual LNCS (or whatever) volume and page numbers.

## Pay Attention to Different Page Sizes

The USA uses 8.5x11 inch paper, while the rest of the world uses A4, which is taller and narrower. Make sure the image area of your paper is fully visible on both sizes of paper (you can check this with Ghostview)--after all, you do want people to read all those words it took such effort to write. The most common problem is papers formatted for A4 that lose something at top or bottom (often including the page numbers!) when printed on US paper. Change the \textheight if this is a problem. Some A4 papers just crash US printers (e.g., high-end HP Laserjets). If you cannot reformat the paper, you can shrink it with the command "psresize -Pa4 -pletter", using the psresize program from the psutils package, which you can get from http://www.dcs.ed.ac.uk/home/ajcd/psutils/

## Sundry LaTeX Tips

I strongly recommend use of the style option cite.sty. This sorts and compresses citations so that [12,10,14,13] becomes [10,12-14] and also allows line breaks at the commas.

If you include URLs in your paper, then use the style option url.sty. This allows line breaks at the slashes and makes things look much nicer.

You can find CTAN (TeX arhive) sites and mirrors by

finger ctan@ftp.dante.de or finger ctan@ftp.tex.ac.uk

## Prepare a Decent DVI and/or Postscript File

Most people with workstations and decent laser printers prefer postscript files, but I've found that many people with PCs and non-postscript printers are grateful if dvi files are available as well (though these are difficult to make self-contained if any non-dvi files were used for figures). 600 DPI postscript files look better than 300 DPI on printers that are capable of that resolution, but the files are much bigger, and take ages to print on modest postscript printers (e.g., Laserjet 4ML), and also slow down Ghostview. Most 300 DPI printers use some form of resolution enhancement that makes the output look close to 600 DPI. Overall, I think it's best to post files at 300 DPI and to save 600 DPI for the camera-ready copy you send to the publisher.

Make sure your postscript file is generated in ascending page order (most printers stack face down, even if yours doesn't) and includes the information that allows Ghostview to find specific pages. (Fire up Ghostview on the .ps file and check you can go directly to page 6, say. If not, get a dvips guru to help you). Make sure the beginning of the postscript file looks something like this:

%!PS-Adobe-2.0
%%Creator: This is dvips, version 5.3 (C) 1986-90 Radical Eye Software
%%Title: testps.dvi

That is, it has line breaks and looks like a normal ascii file, and has the correct %!PS-Adobe-2.0 at the front. Stuff generated with Framemaker generally gets this wrong, and stuff from Windows and Macs often has a ^D stuck in there. Files moved back and forth between PCs and workstations often lose the line breaks. You'll have to edit these bugs out by hand.

## PDF Files

Some conferences require Adobe Acrobat (PDF) files. The ps2pdf script (part of ghostview) can turn postscript into PDF. You'll get much better results if your LaTeX file is set up to use the Times Roman fonts that are built in to postscript and acrobat. Just include \usepackage{times} in the prelude.

There is a better alternative, now: use pdflatex. If you include the following in the prelude, the cross-references and citations become "live" and clickable in acroread. The url package similarly makes urls clickable. The includegraphics command for the graphicx package can automatically choose between eps and pdf graphics provided you omit the filename extension. The epstopdf program can convert eps to pdf.

\usepackage{tocbibind}
\usepackage[bookmarks=true,hyperindex=true,colorlinks=true,backref=true,pagebackref=true]{hyperref}


Note: The new CSL database-driven Web Site automatically takes care of the following topics; I've kept the information here for those who have papers stored outside the centrally administered system.

## Indirect Through an HTML File

If you advertise a URL that points directly to the postscript copy of your paper, you lose all opportunity to revise and embellish the information provided (short of rewriting the paper). It's usually much better to publish a URL that points to an HTML file and have that point to the postscript file. The HTML file can also provide an abstract for the paper that allows people to decide how interested they are before committing to a big download, and allows you to provide pointers to both dvi and postscript versions of the paper.

In addition, an HTML file will get indexed by Google and other search engines, making it possible for people to discover your paper without knowing its URL beforehand. (The awesome Citeseer service indexes postscript files.)

You can also add a BibTeX entry for the paper to the HTML file to make it easier for people to cite it correctly (this doesn't remove the need to have the bibliographic information on the paper itself, since that may get passed around in printed form), and other information such as pointers to specification files, related work, retractions etc.

An example is http://www.csl.sri.com/papers/csl-95-12/

## Keep the URL Stable

Once you post a file on the web, search engines will index it, and other people will make links to it (if you are lucky!). It is vital, therefore, that the URL remains stable--nothing is more indicative of an organization that doesn't understand the web than "404 Not found" messages. This means that not only must you never move a file once you've made it available but that you need to make sure your systems guys don't go changing servers and file systems without placing the necessary redirections in the server. I recommend running this link checker periodically to find dead links.
You might also find The GNU/FSF Web Site Guidelines useful reading.

John Rushby (Rushby@csl.sri.com)