On-the-fly graph generation is useful for dynamic content web sites and automatic report generation. Ploticus is suitable for these situations since it is a non-interactive command line tool with a relatively small code size.

Prefabs may be used, or your plots can be script-based. With scripts, variables are often passed in via the command line. It is also possible to have your program build a ploticus script on the fly, but as this can get fairly convoluted it is recommended only when the necessary flexibility cannot be gotten from passing in variables, embedded #if / #else, etc.

Invoking pl via the shell

This is appropriate if you are using a dynamic web content system such as PHP, or invoking pl via cron or some other automatic command issuer. Generally the procedure is:

1. generate a unique temp file name for the output

2. invoke pl via the shell, using the temp file as the output file (-o)

3. display the temp file using an <img> tag

4. remove the temp file or set up an automatic process to do so. For example, on unix systems you can place the following command into a crontab and run it every night to automatically remove files that are over one day old:
/usr/bin/find tmpdir ! -type d -mtime +1 -exec rm -f {} \;

Direct CGI mode

Another option for producing dynamic web content is to invoke ploticus in direct cgi mode. Within an <img> tag, specify a URL that invokes ploticus with the desired parameters. Advantages over shell mode: you do not need to be running within a dynamic content system (eg. PHP), and shell overhead and temp file I/O are avoided. Disadvantages: since pl is directly invoked as a CGI there are security issues that must be fully understood by the programmer.

From within perl, python, PHP, java, etc.

I currently know of no interfaces that allow ploticus to be called within these environments. Invocation via the shell is recommended.