Now I hear about WWW, and perhaps I should scrap my program, but where do I begin?My program does some complicated assembly of the necessary queries and can present dynamic lists to the user depending on their previous queries, so a simple driver that dump the user into a SQL interface is not sufficient."
The first thing to do is to deal with the user state. That is, where curreently the user wanders through the data in a sort of flow chart, your W3 server will be stateless. The state will in fact be held by the client as the name of the current document. Each time the user does something, the server will get the document id sent over (possible with query text) and will return the text to be displayed and the links to other states.
You have therefore to make an ascii name for his position at any point
in the flowchart. For example, if he has just selected database 1 and has already decided he wants only information available since 1990, then you might sumarize his posistion as something like
/db=1/since=1990etc etc --- plug your own parameters in here. Basically, you are turning the flowchart into a finite state machine here. The state names become the names of the virtual documents which the user will see when he is at that step.Having named each point in the flowchart (there will probably be an infinite number of these as the parameters take arbitrary values) then you
will find that your program consists of code to, given a current state, output the appropriate menu or whatever for the next step. All you do is make a big case/switch/if etc statement to determine which pit of code to run given the state. You have to rewrite the output code so that instead of formatting stuff for the screen directly (building a menu, or moving the cursor etc), it outputs the display in a logically marked up format called HTML. This is like but not the same as say LaTeX. Where a bit of menu
will lead to another state, you generate markup to make it point to
See also: multiple selectionsTim BL