It's not the virtues, stupid ; or, more D-Conf fun

Waldo blogged about the virtues of a common configuration system.

he made some good points, like how it's a configuration system, not a format. yes, there is a distinction and i chose the wrong word in my haste. my bad. he also notes that a standard system for configuration is a good idea, something i agree on now as i did when i wrote my blog.

but he decided to ignore the actual issues i raised and simply recite the happy song of "we like standards". as such i'm highly dissapointed; Waldo simply dismissed the uncomfortable points instead of addressing them.

first, he says that Havoc and he are on the list and they are principles behind gconf and kconfig so that's Good Enough(tm) for having people writing these systems involved. how myopic.

there's a broad world outside of gconf and kconfig. i mentioned Samba for a very good reason. it's a desktop centric application, they have worked up a new backend for configuration and they are itching to work with us. there are also a number of people who have worked on configuration in KDE and there are people working on it now. i named one of them in my last blog, as a matter of fact.

saying "by all means, join the fun" as an invitation to get people in the conversation completely misses the whole point of "extending outreach". this is the #1 reason why FD.o has 10x the noise it should and faces 10x the opposition from developers i talk to than it should: FD.o represents a self selected group of people though it is attempting to (and often successfully) standardizing other people's work and projects. if you don't see how broken that is, i suggest going for a walk in the wonderful spring weather (or fall, if you are in the South half of our planet) and pondering that point.

i'd also love to hear why Waldo thinks FreeDesktop.org should be in the business of software development versus standardization. i don't like software written by committee and i have little faith in the ability for software written specifically for standardization to be as good as software written that is written to be innovative.

i appreciate the time Waldo took to respond to my blog, but maybe next time he'll actually respond instead of deflect. i know he's really into making standards, and i think that's an amazingly critical goal. but we ought to do it right so as not to be left with a desktop platform that's standardized but tastes like cold porridge. because if we do, someone will break from the pack just like has always happened and you can kiss your hard earned standardization goodbye. we had standards on X before this, but they sucked. in part because they were proprietary, and in part because they were designed by committee for standardization not goodness.

we can pretend challenges don't exist or we can deal with them. which will it be?