Oh, sorry, chose to retire entirely voluntarily, only coincidentally after a multi-week settlement process, reports the NY Times in yet another article about itself. This is a big deal, union rules make it tougher to fire a NYT reporter than a tenured faculty member. I like the coy way the article reprints the suggestions and denials of an improper relationship between Miller and Scooter Libby, just so that clued-out readers can know the real reason she's gone. So it turns out that she's not untouchable, but now we know why she was protecting someone who was oh so much more than just a source.
With respect to that relationship, Susie Bright reprints some of Scooter's purple correspondance with Ms. Miller, along with gentle suggestions for improving his next erotic novel. It's sometimes said that everyone has a novel inside them, the real key is figuring out how to keep it in there, and not let it escape into the wild.
Journalists love open source software. Free as both the communist and libertarian senses, David vs. Goliath grassroot movements against giant evil multinationals, all those good storylines. So, naturally, product reviews tend to be at least guardedly positive, don't look a gift horse in the mouth, that sort of thing. Heck, I've even done my bit for OO evangelism.
Now, with the release of OpenOffice 2.0, Massachussetts standardizing on OpenDoc as a document format, a big vague deal with Google, it seems like OO has arrived, and is ready to slay the dominant player in the office software marketplace. Reviews are already starting to be written, no doubt based mostly on Sun's marketing materials, and even those who find the actual To Do list for ongoing development won't find much that sounds like a concern. As a result, there's a serious risk that OpenOffice reviews will be all glowing and nicey-nicey, with perhaps just a couple of boilerplate complaints like "starts up slowly" or "crashes constantly, though usually my document survives". In the vital interest of perspective, then, here's a quick guide to being outrageously negative (yet accurate!) about the all-new OpenOffice.
1. Read what users complain about most, as seen at the OpenOffice.org buglist sorted by votes.
2. Meditate on the fact that entering a bug or a vote into Sun's byzantine and molasses-slow issue tracking system probably requires an hour of work for each first-time complainer, not counting the investment required to isolate the bug in the first place. Each and every bug report was a big deal to somebody. Anything reported in duplicate is huge, a bug with a community; ditto for anything with multiple votes. These are the complaints of the converted, the starry-eyed acolytes desiring to make the world a better place by improving Free Software.
2b. Just in case the starry-eyed acolytes think the System cares, note in passing that Sun's byzantine tracking system 1) hides votes by default, 2) cannot be queried to sort by votes, 3) throws away votes for issues marked as duplicate, and 4) doesn't think this is a bad thing (#22519). That query in point 1 was not easy to create.
3. Goggle at some of the stuff just plain not done. The top complaint, #18285, 500+ votes and ignored patches from a Korean hacker, is this some obscure font thing, of interest only to people needing bizarre languages? That's what I thought. Until I tried to italicize something. That's right, select some text, hit the "I" button, just what the Mac made easy to do 25 years ago, nothing more. Try it with the default font on Linux, it just doesn't work. How about counting words, bugs #4568 / #17964? Sounds simple, and vital for anyone who's publishing, or submitting documents to granting agencies, courts, anyone with space limits (in other words, most anyone writing for a living). But nothing got done at all until someone pointed out that a hypothetical reviewer might try to actually use the app, if only to write their review, and they'd rapidly discover the failing. Now it's a major new feature for 2.0 ("New! with nearly all the functionality of wc circa 1975!") and it's still not right. How about handling notes (#6193)? Pointless glossy frivolity, right? Sure -- unless you have a coauthor, and are trying to communicate about a draft. How about reference tracking, a la LaTeX or EndNote a decade ago, should you be so foolish as to attempt to write your dissertation in OO? Ummm, well there's a whole project devoted to thinking about working on that someday (#4260). Maybe in 3.0. That's all fuzzy-wuzzy humanities stuff though, OO is made for serious technical users, right? These folks are well-served... until they write an equation, and discover that each equation must be vertically aligned by hand (#972) -- couldn't TeX do this 20 years ago? With free, open code, ready to copy or emulate? Or until they try to make a graph in the spreadsheet, and discover that you can't simultaneously plot two independent data series on the same chart (#3997), or have useful error bars (#366). I love the low number on that one. It tells you just how long people have been complaining about it, with nothing getting done. I'm starting to become convinced that the overlap between developers and users of OO is nearly zero, in contrast to a normal community development project, an impression backed up by the fact that 90%+ of OO developers are full-time professionals, mostly Sun employees. If a feature isn't useful to Sun coders, it's hard for them to see the need. Even coder features (eg #10457, request for UML patterns in Draw) get ignored for years if they're not specifically required inside Sun.
4. Pretend you discovered this stuff all by yourself. You're a software reviewer, after all.