McNealy, Sun, and the FOSS community

I've just posted
a long rant in Groklaw

It's about Sun, how great it was that they released OpenOffice under the LGPL, also taking account of the subsquent schyzofrenia they suffer that prevents them from being greater for the greater good.

I don't love Schwartz for his sayings, and McNealy coming out honest won't fix it so quick.

The only behaviour that generates trust is consistence, and that is still something I see lacking across Sun's ranks.


Adam Bosworth gets it and then he doesn't

Adam Bosworth is right about the limitations in so-called "Ajax" applications, and about the limitation imposed by today's browsers.

What I don't agree with: "this is all engineering, not religion, and pragmatism should rule"

This is oversimplifying.
This can be anything we want. It can be science and it can be an art too.
Engineers alone didn't design the way text is laid out in pages, or font types, or behavioural psychology.
Engineers can design awful stuff when they don't have humans in mind.
Engineers should be forced to have a look at: www.ideo.com

Pragmatism should rule in Bosworth's team if he wants it.
I know pragmatism is the philosphy that brought the ill spaghetti patchwork that has been building user interfaces with intermixed html + javascript generated from J2EE apis and the like.

I know those who want to lead and innovate with their users in mind will have the great reward of gaining mindshare and enabling for a future where awful user interfaces is a nightmare from the past.

In this regard, Firefox and Opera including SVG rendering are leading the path towards better web rendering and user interfaces.

John Udell on ActiveGrid

Jon Udell talks about ActiveGrid (a clustering LAMP stack).
Linux: OK, Apache: mmmm... OK, Mysql: cough, cough !!, PHP: ewwwwwwwww !!!

Nice marketing pitch. Get your favourite today's acronym, spice it up with the word "grid", mix it all together, and serve cool ! (though these clusters believe me have little to do with actual gridding !!)
The idea is not bad at all, the problem is not in the scalability of the RUNTIME infrastructure, that can be bearable, can take a little work to make everything work, but it's still doable.

The problem with this approach is that PHP (and please excuse me PHP programmers, but it's the cruel truth) is

1) not a tidy language (doesn't encourage modularisation, reutilisation, namespaces, packages, components, nein, nothing, nada)

2) only recently implemented object orientation, expect quirks and pains until a few minor versions more, still inheriting a shitload of awful procedural legacy, often incoherent, inconsistent function naming

3) against urban legends, it's not fast at all, check the benchmarks against Perl, Python and Ruby.

Mysql seems on the way to be fixed, but why not using other readily available, stable and feature complete databases ? Firebird, Postgres and Ingres have had this for years, if not decades now.

The ActiveGrid concept is good, the problem is this infrastructure doesn't live up to the expectations. Nothing near .NET or J2EE.
For an equivalent, more promising stack, I'd look at www.spikesource.com instead.



This is a test post from flickr, a fancy photo sharing thing.