Question

I wrote a simple webform that will take the user’s information (via POST) and insert it into a database. This isn’t my typical type of programming, but it was basic enough that I didn’t have any trouble.

Anyway, the web form exists in an html page, and is pretty standard. Something like the following (simplified example):

<form method="post" action="/cgi-bin/submit_form" enctype="x-www-form-encoded" id="frm">
   <input id="txtName" type="text" />
   <input id="txtPhone" type="text" />   
   <input id="btnSubmit" type="submit" value="Submit" />
</form>

And I have a C++ application (cgi-bin/submit_form) that runs when it is submitted and stores the data in a database.

I wrote a very basic CGI library that essentially parses the output, which looks something like this:

txtName=<name>&txtPhone=<phone>&btnSubmit=Submit

And the library parses it into a map<string, string> like this:

myMap["txtName"] = "John";
myMap["txtPhone"] = "1.800.555.5555";

Which is then pretty easy to put in the database. However, it is quite evident that my application is completely dependent on the html element variable names (txtName, txtPhone). If any of those are changed, my application suddenly stops working.

---

My question is: How can you synchronize the process of getting data from CGI into a server-side application?

I am very lucky that I’m in charge of both the html and the server-script. But there must be scenarios where one person manages the website and another person manages the scripting, and they must coordinate somehow to ensure nothing bad happens.

In these types of situations, an ICD may be used. For example, when passing XML data, one might use an XSD to define the data structure.

Is there anything like this with CGI programming? I am very frightened at how fragile and easy to break my application is.

Answer 1

You’re referring to the changes between the HTML element names versus the fields being parsed by your CGI script?

As far as something so formal as XSD, I don’t believe so.

Generally, this is the same class of solutions as

• the Unix shell command-line system;
• Environment variables;
• HTTP and SMTP header fields;

… i.e. the POST values are completely free-form strings, and up to your server to read, ignore, or complain about, at-will. (Just as you can freely include “garbage” in your environment, or in SMTP headers, up to the limits of tolerance of whatever system you’re sending them to.)

That said, if you do have a separate designer writing pretty HTML documents and another person writing the application to which they post, there are any number of ad-hoc ways to collaborate. Personally, I’ve generally used Perl’s POD or Javadoc (or whatever the appropriate/analogous inline documentation tool is for your language) to document accepted parameters and limits, and made those available to the art department (or whomever) as nice HTML pages they can pretend to read.

If you were, rather, meaning the SQL table columns, there are any number of tools that will generate fairly decent documentation directly from the SQL server, from SHOW CREATE TABLE foo upwards… The SQL DDL is, itself, a schema on the order of XSD.