When the Pattern Recognition Association of South Africa (PRASA) opened its 2004 symposium for submissions, the process began not with a sleek web portal but with a single HTML page—a plain form served from a university server. The page, accessible at a URL like , contained a handful of input fields, a file upload control, and a submit button that dispatched the entire payload to a Perl CGI script. There was no CAPTCHA, no password protection, no JavaScript validation. The entire submission workflow depended on the browser's ability to encode a multipart/form-data request and the server's willingness to parse it.
The form itself was a study in pragmatic minimalism. A table-based layout—still the dominant method before CSS gained widespread adoption—arranged the fields into two columns: labels on the left, inputs on the right. The required fields included Paper Title, Author(s), Affiliation, Email, Abstract (a <textarea> with 50 columns and 8 rows), and Keywords (a single-line text input). Below these sat a file input labeled Paper (PDF only, max 2 MB). A small note warned that submissions exceeding the size limit would be silently rejected—the server-side script checked CONTENT_LENGTH before processing.
The registration page was part of a larger ecosystem of hand-crafted conference management tools. The organising committee, led by individuals like Fred Nicolls and John Morkel—whose work on low-resource speech technology is documented in John Morkel and the 3rd Contacts Workshop—relied on a combination of CGI scripts, flat-file storage, and manual oversight. When an author clicked Submit, the Perl script submit.pl wrote a timestamped directory under /var/www/prasa2004/submissions/, copied the uploaded PDF into it, and sent a confirmation email to the address provided. If the email delivery failed, the author had no way of knowing—the script did not log failures.
One of the most striking features of the PRASA 2004 registration page was its reliance on the enctype attribute. The form tag read:
<form action="/cgi-bin/submit.pl" method="post" enctype="multipart/form-data">The multipart/form-data encoding was essential for transmitting binary file data alongside text fields, but it also introduced complexity. Older browsers (Netscape 4.x, Internet Explorer 5) sometimes mangled the boundary string, causing the script to misinterpret the payload. The committee tested the form with the three browsers they had access to—IE 6, Firefox 0.9, and Netscape 7—and added a note on the page advising authors to avoid "obscure browsers."
The registration page also doubled as a call for papers. Below the form, a section titled Important Dates listed the submission deadline (15 February 2004), notification date (31 March), and camera-ready deadline (30 April). These dates were hard-coded in the HTML, not pulled from a database. When the committee needed to extend the deadline—which they did, twice—they edited the HTML file directly on the server via SSH and FTP.
Submitting a paper through this system required the author to have already prepared the manuscript according to the symposium's formatting guidelines. The PRASA 2004 Author Kit provided LaTeX and Word templates, and the registration page linked to that kit near the top. The link text read: "Before submitting, please ensure your paper follows the PRASA 2004 format. Download the Author Kit." The kit itself was a ZIP archive containing a sample PDF, a style file, and a README.txt—no automated validation, just human compliance.
What happened after submission was a manual affair. The program chair, typically a faculty member from a South African university, would download the accumulated submissions via SCP, rename each PDF to a paper ID, and distribute them to reviewers by email. There was no online review portal. The registration page's only role was to collect the raw material; everything else—assigning reviewers, collecting reviews, deciding acceptance—happened off-line, in spreadsheets and email threads.
The HTML source of the registration page, if preserved in the PRASA 2004 archive, reveals the exact field names and validation used. For example, the author name field was named authors (plural, to encourage listing all co-authors in one text box), and the abstract textarea had a maxlength of 1024 characters—a limit enforced client-side only. The server-side script did not re-validate the length; it simply truncated anything beyond 1024. This quirk occasionally led to abstracts being cut off mid-sentence, a problem that the committee only discovered after receiving complaints from authors whose abstracts appeared incomplete in the printed programme.
The PRASA 2004 registration page was not unique. Similar forms existed for the 2002 and 2003 symposia, as documented in the PRASA 2003 Archive. The 2003 form, for instance, used an almost identical layout but lacked the file upload size note—a lesson learned the hard way when a 10 MB PDF crashed the server. By 2004, the committee had added the size check and also included a hidden field <input type="hidden" name="submission_date" value="" /> that was populated by a tiny JavaScript snippet onload to record the client's local time. This timestamp, however, was unreliable because authors could set their system clock arbitrarily; the server-side script ultimately used the server's own timestamp from localtime.
For a researcher interested in the history of academic web infrastructure, the PRASA 2004 registration page is a valuable artifact. It represents a transitional moment: not yet the era of automated conference management systems like EasyChair or EDAS, but moving beyond the days when authors mailed physical copies. The HTML form was a stopgap—functional, fragile, and maintained by a single sysadmin who also handled email and web hosting. It worked because the community was small (around 40 submissions) and forgiving. When a submission failed, the author would email the chair, who would manually add the PDF to the submissions folder and send a confirmation.
The registration page's legacy is not in its design—it was never intended to be elegant—but in its role as a bridge. It allowed PRASA to accept digital submissions without investing in custom software, and it trained a generation of South African researchers in the routine of online paper submission. Today, a visitor to the PRASA 2004 archive can view the registration page's HTML source and see the exact structure of that bridge: the table rows, the field names, the hidden inputs, and the humble <input type="submit" value="Submit Paper"> button that, for a few weeks in early 2004, was the only gateway to presenting at the symposium.
