Greetings, all,

A quick update on the hackyInstaller integration process:

* Austen and Julie have been working with Mike to create a stand-alone version of hackyInstaller (in the module hackyInstaller2) that satisfies the requirements of the UMD group for batch, multi-user sensor installation. As of 3am this morning, Austen appears to fixed all of the known issues.

* Julie and I have been working on an integrated version of hackyInstaller. By "integrated", I mean one that gets built automatically as part of the hackystat build process (similar to the way the sensorshell gets built.). We have been focusing on building a 'minimal' instance of hackyInstaller with only one sensor isntaller definition. Thanks to a lot of heavy lifting by Julie, I can now do a successful integrated build in my local workspace. I'm now fiddling with the daily build site to get it to work there as well. I'm hopeful that will happen today.

That puts us in the following situation:

1. Mike needs to do acceptance testing on the current version in hackyInstaller2 and check to see that it satisfies the requirements. He can then release it to UMD for trial use.

2. Austen needs to merge the changes from hackyInstaller2 into the regular hackystat sources. This is an incredibly error-prone task, so I suggest you work with Julie to make sure you integrate everything necessary, build a local copy of the integrated hackyInstaller, and test to make sure all of your fixes have been merged.

3. Julie, Austen, and James can now go ahead and migrate the remaining sensor installer definitions from hackyInstaller2 to the appropriate hackystat cvs modules (off the top of my head, these modules probably include hackyAnt, hackyEmacs, hackyVim, hackyJBuilder, hackyEclipse, hackyVCS, hackyJupiter, hackyCLI, hackyOffice, hackyPerf, and hackyVisualStudio). Several of these modules do not currently have any java code in them, so you will need to import the appropriate directory hierarchy (org/hackystat/etc etc.). All of these modules will need their local.build.xml file fixed to make sure that the <compile> task compiles the sensor installer code and the <install> task copies the appropriate installer.*.xml file over to the appropriate location for the build.

4. I will start revising the client sensor installation chapter of the docbook to document how to use hackyInstaller for installation purposes.

5. We make some releases to the public server (hopefully starting today) to test stability.

6. Once it's all done and everything looks OK, we'll make a stable release of 6.7 with hackyInstaller.

This is almost certainly going to be the first stable release of hackystat that cracks 100,000 LOC. Shall I get a cake or something? :-)

Cheers,
Philip

Reply via email to