The Complete Magazine on Open Source

Dryrun: Decoupling the Process

547 0

Dryrun. Come again?

We take a look at Dryrun, an email-based interface via which you can run programs.

Dryrun attempts to present an email-based interface to run your programs. For example, if [email protected] is the designated email address, an email message such as the one below will result in a reply to the sender with a message containing the output of the submitted code.

To: [email protected]
Subject: python

print "Hello World!"

The reply will contain a file, process.log, which has the following message:

Output produced during processing of your file
Hello World!

This is a simple example of how dryrun works. So why would this be useful to you? The usefulness of this idea is debatable. However, what it allows you to do is use any Internet-connected device to send code snippets in an email, and get back the output. As you think of ways in which you may plan to use such a service, let us next look at how you can run your service.

Running your own dryrun service

For dryrun to function, you need to make sure that the code is running to receive requests. The source code consists of three Python files:


This is the driver script for, documented separately. This script first invokes dryrun_invoke() to check for any existing unprocessed requests, then sets the IMAP server in IDLE mode, so as to be notified when a new email arrives.

It is built on the code from here. You will need to specify the details of the SMTP and IMAP server you plan to use, in the global variables SMTP_SERVER, IMAP_SERVER, username and password.

This program implements the core logic of the program. The entry point to this program is the dryrun_invoke method, which is called by the script. In this script, the imap_access method is used to obtain the request details.

Then, it checks if the programming language is supported, using the ‘extensions’ dictionary. If supported, the execute_code method is called to run it. Here, the appropriate program to process the code is chosen: gcc for C, g++ for C++, and python or pdflatex for LaTex files. The Python subprocess module is then used to run this code after it has been written to a file named code.<extension>, where the extension is chosen appropriately from the extensions dictionary.

The output obtained is then stored to a file called process.log, which is sent back as an email to the sender, in the method send_output.

This is the threaded IMAP4 client based on RFC 3501 and the original imaplib module. To run this service, simply execute, using python Ensure the appropriate programming tools are installed, and try out how it works by sending an email in the form discussed above.

Including your language

Of course, it’s very simple to extend this to include a language of your choice. As of now, you can directly modify the ‘extensions’ and ‘program’ dictionaries in the file However, you will have to manually include the logic your particular language may need, before you get the output. For example:

  • Python code execution is a one-step process
  • C/C++ is a two-step process: gcc/g++ and then ./a.out
  • LaTex produces PDF output, so it needs to be attached separately, along with process.log.

Where to next?

You could use this email-based service as a backend service and have different frontends, such as a bot, a Web-based interface, a smartphone application, etc. Again, it is a project that I wrote to have fun, and I guess that’s where its usefulness ends. Let me know what you think!