Application Development With Webware

Webware for Python




Here we describe best practices for developing a web application using Webware.

Setting up your application

The first task in developing an application is to set up the file structure in which you will be working.

It is possible to put your application in a subdirectory under WebKit/ and change WebKit/Configs/Application.config to add another context. But do not do this. Your application will be entwined with the Webware installation, making it difficult to upgrade Webware, and difficult to identify your own files from Webware files.

Creating a Working Directory

Instead you should use the script bin/ You should run it like:

$ python Webware/bin/MakeAppWorkDir -c Context -l Lib --cvsignore WorkDir

This will create a directory WorkDir that will contain a directory structure for your application. The options are:

-c Context:
Use Context as the name for the application default context. A subdirectory with the same name will be created in the work dir (you can change that with the -d option). If you do not use the -c option, the context name will be MyContext. I like the name Context for all my applications.
-l Lib:
Create a Lib directory in the work dir which will be added to the Python path. You can use the -l option multiple times; and you can also add already existent library directories outside of the work dir.
Create .cvsignore files for use with CVS.
The files will be put here. Name if after your application, place it where it is convenient for you -- it doesn't need to be located close to the Webware installation.

You can see all available options if you run Webware/bin/ without any parameters.

When you do this, you'll see this directory structure:

AppServer*  Configs/  error404.html  Logs/      WebKit.cgi
Cache/      Context/  ErrorMsgs/     Lib/       Sessions/  webkit*

Here's what the files and directories are for:

The script to start up the AppServer for this application. Each application will have its own AppServer, and its own process. If you are running under Windows, you will see a AppServer.bat instead and additionally, you will find a script that can be used to start the AppServer as a service.
A directory containing cache files. You won't need to look in here.
Configuration files for the application. These files are copied from WebKit/Configs, but are specific to this application/AppServer.
The directory for your default context. This is where you put your servlets. you can change its name and location with the `-c and -d options. You can also change this subsequently in the Application.config file in the Configs directory, where you can also configure more than one context. You may also want to remove the other standard contexts that come with Webware from the config file.
The static HTML page to be displayed when a page is not found. You can remove this to display a standard error message, modify the page according to your preferences, or use a custom error servlet instead by setting ErrorPage in the Application.config file appropriately.
HTML pages for any errors that occur. These can pile up and take up considerable size (even just during development), so you'll want to purge these every so often.
Called by the AppServer script to launch the AppServer.
An example for an application-specific library package that can be created -l option (in this case, -l Lib. Import modules from this directory like from Lib.SitePage import SitePage.
Logs of accesses.
Users sessions. These should be cleaned out automatically, you won't have to look in this directory.
A CGI script/adapter for accessing the AppServer here. You can still use the other adapters, but most of them don't need to be configured for the individual applications. I still recommend mod_webkit or wkcgi.
If you are running under Unix, you can use this as a start script (see the WebKit Install Guide).

Using a version control system for your application

A version control system is a useful tool for managing your application. Popular Open Source version control systems are are the Concurrent Versions System (CVS) and, increasingly, Subversion (SVN). I recommend using SVN because it has a few advantages over CVS. For instance, it tracks both files and directories and handles copy, rename, and delete operations on files well. These systems handle versioning, but they also make it possible for other people to see snapshots of your progress, for multiple developers to collaborate and work on an application simultaneously, and they create a sort of implicit file share for your project. Even if you are the only developer on an application, a version control system can be very helpful.

The working directory is a good place to start for creating a versioned project. Assuming you've set up CVS, and set CVSROOT to point to your repository, you can get started by importing your project into the repository simply by running:

$ cd WorkDir
$ cvs import -m 'initial import' MyWebwareProject username start

Replace MyWebwareProject with the name of your project and username with your own user name. You should use the option --cvsignore when running if you plan to do this. If you do, then .cvsignore files will be added to each directory. These tell CVS to ignore files with certain extensions (such as .pyc files), and all the files in certain directories (Cache, ErrorMsgs, Logs, and Sessions). You shouldn't otherwise notice these files, even if you aren't using CVS.

The command to import your project into a SVN repository is very similar:

$ cd WorkDir
$ svn import  -m 'initial import' https://myserver/myrepos/MyWebWareProject

Replace https://myserver/myrepos/ with the URL of your SVN repository. The .cvsignore files will not be used in this case. Instead, you have to set the svn:ignore property on the respective directory. You can still use the .cvsignore files to generate the necessary svn propset commands:

$ find . -name .cvsignore | while read f; \
>   do echo svn propset svn:ignore -F $f $(dirname $f); done

After importing WorkDir to the repository, note that it is not automatically under version control. To start working, you first need to explicitely check it out from the repository using cvs checkout or svn checkout.

Using the working directory from multiple accounts

If you are using a version control system or if you are otherwise distributing your application code, you may find that it is difficult to manage the differences between accounts. For instance, in different accounts on different machines Webware may be installed in different locations. You may have the actual directory in a different location as well -- it may be in ~/webware/WorkDir for your active development, but /var/webware/WorkDir for the production version. And if there are multiple development copies on the same machine, you have to be sure they each use different adapter ports.

To solve these problems I recommend creating a shell script to handle startup. I generally call this script start, and it looks something like this:


# is my development machine
if [ `hostname` = ] ; then

# this is my production environment
if [ `hostname` = && `whoami` = webware ] ; then

if [ "$WORKING" = "" ] ; then
    echo I do not recognize this environment
    exit 1

./AppServer --work-dir=$WORKING --webware-dir=$WEBWARE $OPS $*

You can add this to your project in the repository, and the script should automatically detect what environment it is being used in. You can use options to change configuration parameters, like setting some parameters depending on whether the environment is a development or production environment.

Some options that you may be particularly interested in:

Setting this to 1 will make the AppServer restart if there have been changes to any loaded files. This is very nice during development.
If you want multiple applications running on the same machine (e.g., one for development, one for production), you have to use different ports.
You probably don't want to have this on in production, but it's nice during development.
During development you probably want this off.
Turn on for production.

For more settings, see the Configuration document.

Structuring your Code

Once you've got the basic files and directories in place, you're ready to go in and write some code. Don't let this document get in the way of developing the application how you choose, but here are some common patterns that have proven useful for Webware applications.


Subclass a SitePage from WebKit.Page for your application. This subclass will change some methods and add some new methods. It serves as the basis and as a template for all the pages that follow.

Some code you may wish to include in your SitePage:

  • Authentication and security
  • Accessing common objects (e.g., a user object, or a document object)
  • Page header and footer
  • Common layout commands, like writeHeader
  • Database access

I also typically add other functions to the SitePage module, and then do from Lib.SitePage import * in each servlet -- this might include functions like htmlEncode, or some other select functions that I use constantly in web applications. Whether you want to use functions or methods is up to you -- in many cases methods can be more easily extended or customized later, but sometimes method use can become excessive and create unnecessary dependences in your code.

A basic framework for your SitePage might be:

from WebKit.Page import Page

class SitePage(Page):

    def respond(self, trans):
        if self.securePage():
            if not self.session().value('username', False):

    def securePage(self):
        """Override this method in your servlets to return True if the
        page should only be accessible to logged-in users -- by default
        pages are publically viewable"""
        return False

    def respondLogin(self):
        # Here we should deal with logging in...

Obviously there are a lot of details to add in on your own which are specific to your application and the security and user model you are using.