Agylen

It might be a bit early to say something interesting about krugle, given that it hasn’t launched yet, but I’ll try.

Judging from the product brief alone, it should really rock. But of course, all that matters is execution. I think relevance can be a big issue here. One of the screenshots shows search results for “java hash map”. How do you decide that a search result from Eclipse is more relevant than a search result from Apache Portals?

Koders has the same problem here, though. However, searching for “java hash map” is not very meaningful: Any competent Java programmer would just use a java.util.HashMap. It would be more useful to search for, say, “red-black tree algorithm in java”. There Koders fails miserably, whereas Google does find some results which seem relevant. How will Krugle fare? I don’t know, but if I manage to be accepted in the beta program, I will surely report.

I have also some reservations about the emphasis on componentized software:

The world has shifted from inflexible, monolithic applications written in isolation to applications built with components that can easily be modified and leveraged.

Unfortunately, at least in my humble opinion, the promise of reusable, modifiable components has failed to materialize so far, except maybe for VB programmers. It’s true that you can find many freely available libraries, toolkits and frameworks, but try to find an application-level component that you can readily reuse without extensive rewriting and you’re suddenly faced with a much harder task.

Anyway, this shouldn’t detract from the fact that a search engine for Open Source code can actually be useful, if properly implemented. I’m currently thinking whether it would be possible and worthwhile to mash up The Open Source Zone with Krugle. Do they provide an API?

Last but not the least, they have Chris Locke on board as “Blog Wrangler”. Well, this probably shows that they have a clue, at least.

Technorati Tags: open source, krugle, chris locke

I just finished adding a new feature to The Open Source Zone, inspired by a similar feature on CodeZoo. Users can now add short “tips” to a project. The tip can also contain a snippet of source code. Hope you’ll appreciate it.

Here’s a new feature for The Open Source Zone: A nice monthly calendar for Open Source conferences.

Head over to The Open Source Zone to check it out. And if your favorite conference, convention or symposium isn’t there yet, feel free to add it yourself.

Of course, you can always use the Conferences RSS feed to stay up to date on new conference announcements, just like before.

Sacha Labourey: “As I was thinking about the way companies (mostly software vendors) position towards Open Source, I realized I could try to categorize them.

Here is what I came up with:

• The truly committed
• The mixed-codebase
• The pragmatics
• The anti-strategist
• The headless chickens
• The in-denial
• The anti-OSS”

I like to think that we can be characterized as mixed-codebase:

The second category, the “mixed-codebase”, gathers companies with an increasing commitment to an Open Source strategy but that also have some existing critical revenue streams to protect, and hence cannot simply switch to a full OSS strategy overnight. “Mixed-codebase” companies not only deploy on/use Open Source software (as an ISV), but also actively contribute to (or even lead) OSS projects and have some clear OSS alliances reinforcing this strategy. Examples of such companies include Novell and Computer Associates. I mostly see the “mixed-codebase” category has a transition step: companies will eventually move to category 1 (successful transition) or 3 (retrenchment), depending on how well they execute their transition, how well they can convert their existing revenue streams, how frequently they change their CEO, etc. I do not know of any company that would have successfully switched from a traditional company to a category 1 company yet. My bet is that such a thing will take place in the next 2 years.

We are probably also trying to move to category 1, but it’s going to take lots of time and sweat. Where are you on the above scale?

I just found out that the “smart” tagging interface I implemented last friday is not so smart after all. It’s actually a bit broken in that it only extracts single words from descriptions before matching them with existing tags. The problem is than tags can be made up of multiple words and no matches with multi-word tags are thus ever found.

Alright, this can be fixed and I will probably work on it tonight, but this made me think about the reasons why del.icio.us tags do not contain any spaces. I still think it’s inconvenient, from a user’s perspective, not to be able to use spaces to separate words within tags and to be forced to use hyphens, underscores or other similarly awkward punctuation marks. But I can understand how easier it is to deal with it form a programmer’s perspective.

Another problem is that it will be impossible to connect OSZone tags to del.icio.us tags without arbitrarily changing blanks into underscores or something like that, before linking.

Anyway, users always come first, so I’ll stick to spaces for now. Changing along the road, if it is necessary, is just a search & replace operation.

I also started thinking about the taxonomy of projects. Until now, we have relied on a simple system of categories that is intended to classify projects according to their main mode of use: libraries that you link in with your code as opposed to frameworks that somehow force your code to adapt and be controlled from the outside, or development tools that you use during development but don’t become part of your program.

I reasoned that tags, the search and the “similar projects” feature would be much more useful than putting projects into a fixed set of non-communicating vases. So I went for the minimum amount of taxonomy that would not get in the way too much. I still think this is a good idea, but I’d like to explore adding another dimension to the classification.

This would be the dimension of “field of application” and its axis would be labeled “web publishing”, “data management”, “XML”, “graphics”, “text manipulation”, … (I’m just picking categories at random here, out of the top of my head). I would never want a hierarchical taxonomy, but probably a set of orthogonal dimensions along which to classify projects would be nice to have. User could choose to navigate along the dimension that they care the most about. Let’s call that faceted navigation (hat tip: the Daisy folks). What do you think?