In the beginning there was server-side rendering, providing online math as a Web service. That meant that the \LaTeX\ was sent from the client to the server, where image files were generated and then served back to the client for display within an {\tt img} tag in an HTML document. 


I attempted this quite some years ago with some success. I utilized the ``textogif'' command on a server and enabled real-time AJAX communication to dynamically serve generated images. The results were similar to that of MathTRAN, another server-side rendering solution. 


A major downside of this is that the server has to generate files and serve them, and the client must make HTTP requests for every single equation, which is very inefficient. 


Another issue with MathTRAN was that the syntax was verbose. You had to set an HTML element's attribute to the \LaTeX\ code you wanted rendered, so it was a very disconnected editing experience, since you had to combine HTML and \TeX.


MathTRAN was some sort of \TeX\ and HTML:

\begin{print}
\begin{verbatim}

 <p>The equation of the unit circle <img alt="tex:x^2+y^2=1" /> and
   a related trignometric identity <img alt="tex:\sin^2\theta +
   \cos^2\theta = 1" />.  </p>
   
\end{verbatim}
\end{print}


\begin{interactive}
\begin{verbatim}
   
 &lt;p&gt;The equation of the unit circle &lt;img alt=&quot;tex:x^2+y^2=1&quot; /&gt; and
   a related trignometric identity &lt;img alt=&quot;tex:sin^2theta +
   cos^2theta = 1&quot; /&gt;.  &lt;/p&gt;
   
\end{verbatim}
\end{interactive}


This type of authorship interface cannot put an author into a ``flow''. HTML, as much as I love it, is starting to resemble a sort of assembly language.

Server Side Rendering