Kimbro Staken wrote:
On Dec 4, 2003, at 7:58 PM, Murray Altheim wrote:
Ryan Hoegg wrote:
I hate to be a wet noodle or a wet blanket (or, now living in England, anything wet), but I've been watching this project since its days as dbXML, and while there are now and then a few people who pipe up as interested, and even rarer a few who contribute time and code, I look at that mythical number of 747,939 registered SourceForge users and wonder if there are truly businesses and people banking on native XML databases, how we can't seem to locate even a handful willing and able to put time into Xindice. Unless everyone is going to commercial DB, which is hard to imagine.
Well, I don't think it is that simple. Databases are an entirely different class of project over the typical Sourceforge effort and are intimidating to a lot developers. Rightfully so too. The Xindice project has always been able to find people to work on the periphery, but we've never been able to fill in the hole around the database core created when Tom was forced to move on. Having that hole kills the progress of the project overall as there's no momentum on any major issues and the people drift away. I personally tried to fill that hole, but I've never considered myself a "hardcore" programmer and that was proven by my lack of progress in the area.
Well I can't speak for other people, but whenever I get involved with a project, especially one that's been around for some time, the reason for the initial involvement is to scratch an itch. That often means getting involved with things on the periphery at first: documentation, tools, maybe easy optimizations, and mostly fixing the few cases that don't work for me for some reason, possibly because others haven't encountered the same conditions yet. Frankly I'd be worried if I jumped in out of nowhere to rewrite the core code completely and the people in charge let me. One has to earn some level of trust before that, and while that goes on one gets more familiar with the product which is a Good Thing.
So that's exactly what I did. I installed Xindice. I tried it. I tested it. I started using it in some projects I was working on that could benefit from a native XML database. In doing so I stumbled upon some limitations of the engine, on cases where it started misbehaving and corrupting the database. By that point I was becoming more and more familiar with the engine so I felt more confident about going in to investigate, and that's what I did. I spent lots of time to understand what was going on and figuring out how to solve the problems I encountered. And I believe that was in the core of the database (if the filers aren't core, I don't know what is). When I emerged with the solution, I tested it mercilessly for days, and the tests showed that the problems were gone and everything else still worked flawlessly. But I couldn't just commit my changes, as I don't have committer status. So I did what the documentation said I should do in that case: submit a patch to bugzilla. I also sent a message to this mailing list. The bug has remained unopened in Bugzilla to this day.
That's how you lose the interest of competent potential contributors. I don't have the resources at this point to maintain a fork of Xindice. It makes sense to me to contribute part of my time to the project so it can improve for everyone (including me). It doesn't make sense to maintain my own Xindice on my own. If the powers that be have moved on to other things (which is their right) but are not doing what it takes for the project to go on living (which is sad -- none of the committers have committed anything for what, months?), then what's happening now is inevitable: some great people notice Xindice, try it out, try to make it work, want to contribute and are ignored, and then they move on.
The thought of having the project die is very disheartening to me, as Tom mentioned, we both really suffered from our desire to continue working on Xindice. Unfortunately, we have to face the fact that databases in general are hard, and there just aren't a lot of people who can tackle them. Most of the larger Apache projects have people paid to work on them, Xindice hasn't had that luxury.
True, but most of the larger Apache projects started out small too, and had to prove themselves first. And don't forget about the people who are paid to work on other projects that will choose to work with Xindice, who will improve it, and will give that back to the community. Sure we are not paid by the Apache Foundation, but does it matter at all? I know that if I had felt like Xindice was still going somewhere, and my contributions were eventually going to make it into the product so I don't have to maintain my own fork, we would still be using it, I would have contributed lots more improvements and I would still be bringing in even better contributors. And I wouldn't be alone in doing so.
As I said previously, a project lives or dies by having a handful of (or even one!) truly committed and talented people involved. I've got a project of my own right now that has taken up the lion's share of my time (like, night and day) for the past two years, so I do speak from some experience. If Xindice had one or two people similarly committed to its success, it would succeed. It won't simply by having people wanting it, or "banking on it". There needs to be some people *working* on it.
That's true, but... it's hard to have that level of commitment to a very difficult project that you didn't create. This is why I suggested that maybe it should be open for a new start. Whether that can or would succeed I don't know, but without anyone even willing to try it doesn't matter.
I don't believe that's entirely true. Xindice is a very interesting project in that it already has a working base, and has lots of potential and room to grow. People can and will be willing to commit some energy on the project, but they need to feel like their contribution is not wasted on everyone. I sincerely hope that the Xindice project can turn itself around and start to prosper again. I'm encouraged by the amount of response this discussion is having. It certainly is a good sign.
Mario
Murray
...................................................................... Murray Altheim http://kmi.open.ac.uk/people/murray/ Knowledge Media Institute The Open University, Milton Keynes, Bucks, MK7 6AA, UK .
Kimbro Staken Software, Consulting and Writing http://www.xmldatabases.org/ Apache Xindice native XML database http://xml.apache.org/xindice XML:DB Initiative http://www.xmldb.org
