First, I'd just like to introduce myself - if nothing else to help give a background to questions/comments that will crop up from my corner. I'll follow up in another mail with one or more actual questions, so if you're not interested in this intro feel free to skip. Apologies if this seems too lengthy...
'The Frameworx Company', is based in the US (New York) but also has a Swedish subsidiary. I'm myself Swedish but currently located in the US. The development organization is about evenly split between the offices. We have for the last three years labored with developing an advanced Java framework intending to enable building large-scale web sites using J2EE technology. The framework supports a layered architecture from the ground up - raw database, UML-to-information model with object-to-relational mapping and our own OQL, business object layer, application object layer, interaction layer, workflow & workrouter facilities etc etc. All expressed as declarative XML models, interepreted and run by the framework. There is also a Swing-based editor enabling high-level graphical editing of the XML necessary.
However, even though I'm also a core developer, I have been mostly involved in the inner workings to enable the rest of the development organization to function: making sure there is CM that maintains some order. I'm an experienced ClearCase/Multisite administrator; I've worked with it for the last 8 years so that was the natural choice when we started.
Our codebase grew very quick, and we also expanded quickly, getting lots of junior developers that had very little experience with CM systems. So, to avoid 1) ever-increasing build times, and 2) total chaos because people would check in almost anything, it prompted me to create a toolset, implementing a process on top of CC. This has evolved into a large-scale Perl wrapper application that covers a CM process with formalized code deliveries and checks, builds, release management etc. It also implements 'components' as a versioned entity - this enables the tool to substitute binary component versions dynamically instead of source. This helps to substantially lower build times and also promotes good OO design for separation of concern, interfaces vs implementations etc. Eventually, it has evolved into a product in its own right. It's currently transitioning into a more generic Java toolset.
Given the way the industry is moving, we're planning to open-source our framework and other things we have. To make this possible, a ClearCase installation is not realistic...and frankly, I don't think I could live with CVS (directory versioning, a chief concern) - so, I'm planning to use Subversion. I'm impressed by what I see. Even though I'm a little bit apprehensive about the relative immaturity of it, I think it has a very good future ahead of it. Of course, there are still things that I will miss dearly from CC (anyone who has used the graphical version tree for elements will know what I mean), but that will come, I'm sure.
My intention now is to look into how I can map my componentization on top of svn, adapt my current tool to talk to svn, and also how I should set up tracking of configurations (i.e. mixes of components and versions of them). I'm also hoping to be able to contribute to the svn effort once I've got things moving, primarily on the win32 side where I did system level C/C++ work for several years before. Once we're moving I'll announce it here for interested parties.
Thank you all for your time and effort with svn so far,
Senior CM Architect
The Frameworx Company
Everyone else has a cool sig, so I have a feeble one.
To unsubscribe, e-mail: email@example.com
For additional commands, e-mail: firstname.lastname@example.org
Received on Sat Sep 20 07:51:30 2003