Planning

The bare minimum required to get started is a JDK, a JAXP 1.1 compliant parser, a command prompt (bash or Windows NT/2K), and Perl. Perl is only used at the moment for some of the publishing and notification, so it can be omitted. It is highly recommended at this point that you review the description of the data definitions used by Gump, starting with the overview. It also would be useful to gather up a list of installed packages that you will be referencing in your build. A list of some of the packages you might want can be found here.

Getting the source to Gump

export CVSROOT=:pserver:anoncvs@cvs.apache.org:/home/cvspublic cvs login cvs co -d gump jakarta-alexandria/proposal/gump If prompted for the password, enter anoncvs

Defining your workspace

Two sample workspaces are provided, rubix and rubypad. Find the one that is closest to your needs and make a copy of it. The suggested name for the file you create is your host name. Again, reference the overview to see how you can add projects, define your own profiles, and the modify which version of referenced projects you want to build. You will also need to directly modify the gen.bat or gen.sh scripts to specify where to find the JAXP implementation to use. At the present time, recent releases of Xalan are recommended for performance reasons; but the JAXP reference implementation will likely catch up shortly. Note: Yes, it is tacky that you should have to modify this file directly. Patches are welcome. Meanwhile, cvs does a fairly good job of merging changes.

Generation

For Unix and cygwin installations using bash: bash ./gen.sh For Windows NT and 2K installations: gen This process generates the update and build scripts used in subsequent steps. It also logs you onto any :pserver: cvs repository defined which you are not currently logged onto, and generates the various cross reference pages.

Update

For Unix and cygwin installations using bash: bash ./update.sh ( all | module ... ) For Windows NT and 2K installations: update ( all | module ... ) With this command you can specify you either want to update all or a specified list or modules. If the module had not been previously checked out (or the target cvs directory has been removed) then a checkout is done instead. If all is specified then a web page containing the logs for each update is produced, as well as a summary page. If specific modules are listed, then the output comes to the screen.

Build

For Unix and cygwin installations using bash: bash ./build.sh ( all | clean | ( project [ target ... ] ) ) For Windows NT and 2K installations: build ( all | clean | ( project [ target ... ] ) ) With this command you can specify you either want to build all or a specified project. If all is specified then a web page containing the logs for each build is produced, as well as a summary page. If clean is specified then the only action that is performed is to synchronize the build directories from the cvs directories. If a project is specified, then the output comes to the screen, using the target you provide.