A question for tech writers: why does every technical manual and book include a section at the beginning on "conventions used in this document," full of ridiculous and useless tips like, "Tips are indicated by a lightbulb icon in the margin"? Is it because you're paid by the word?
To steal a joke from Mr. Bunny's Guide to ActiveX, I would like to announce that "the familiar dot (.) symbol from Internet addresses shall be used on this website to denote the end of a sentence."
Last year I was designing the upgrade installer for CityDesk. I was considering RTPatch but it was really expensive (disproportionate to what we would spend in bandwidth), so we just made an upgrader that reinstalls full copies of every file that had changed.
At the time RTPatch was $5000 for a license, and their salesman called me up to give me a hard time for contradicting myself. After all, step 9 in the Joel Test is using "the best tools that money can buy." "Touché," I said, and hung up.
Now there's competition. A little Israeli company, Red Bend Software, has a product called vBuild, which is tightly integrated with InstallShield Professional (the non-Windows MSI version of InstallShield). vBuild is about $2500 and RTPatch has come down to $2750. Still pricey but starting to seem worth it. Today I'm going to try converting CityDesk's setup from InnoSetup, which is excellent and free, to InstallShield, which is decent and expensive, just to take advantage of vBuild.
He also pointed out that Wise for Windows Installer now has a ton of features for server setups: user account creation, IIS administration, and SQL Server administration... basically they implemented my entire wish list from working on the FogBUGZ 3.0 installer. Good work!
Some people on the forum expressed concern with the concept of distributing upgrade patches rather than complete copies of changed files. I can see why they are concerned, but I think they are underestimating the tools that are out there. These tools check the version of each file on the user's machine and also compute a hash function to make sure that the file is not corrupted. If the file is corrupted, they automatically download the whole thing.
Using patching technology saves a lot of bandwidth, even for the first-time installer, because a lot of the bytes you're distributing may be system files and runtimes that users already have. For example, we need to redistribute things like the Microsoft C Runtime (MSVCRT) which almost everyone already has, just because there may exist a few Windows machines in the wild which don't have it. For consumer software which may well be downloaded over a modem, this makes a big difference.
For now, Michael convinced me to stick with the tried-and-true Inno Setup for the next release of CityDesk. The bandwidth savings don't quite justify the effort of switching installers.
Chandler Release 0.1 is out. See what all the fuss is about, almost.
Larry Seltzer wants to throw out SMTP and start over. I have to support that. If the European Union can change the money that 300 million people use all on one day, we can change our email protocol. Let's find a protocol with decent authentication and with micropayments to make spam uneconomical, and let's set a deadline, maybe two years in the future, when SMTP will simply be turned off. Everybody will know it's coming, just like Y2K, and everybody will have to be ready.
A couple of new books worth checking out:
Waltzing With Bears: Managing Risk on Software Projects. The latest book from the heroes of Peopleware, Tom Demarco and Timothy Lister. Makes a brave case that the person (or company, or team) who is responsible for risk management is the person who is going to pay the price when things go wrong. Waltzing With Bears kicks off with an excellent analysis of the famous "failure" of software engineering that delayed the opening of the new Denver airport, framing it as a failure of risk management. In fact when you look a bit more closely at the software project itself, it's pretty clear that the software team was expected to perform the impossible, and the project as a whole had no way to recover from a slip in a typically over-optimistic software schedule.
Speed Up Your Site: Web Site Optimization by Andrew King. A well written and very thorough treatise on what you can do to reduce download times and speed up web sites. Yes, putting your whole HTML page on one line can really help.
No-Collar: The Humane Workplace and Its Hidden Costs, by Andrew Ross. A big disappointment. I think the original book was not exciting enough so the publisher tried to present it as a critique of informal, humane workplaces. The "hidden cost," they want us to believe, of pleasant, diverse workplaces seems to be that they shut down and you lose your job. I know, it doesn't make any sense. The author wanted to write a book about cool dotcoms, but they all imploded, and he was stuck trying to claim that they imploded because they tried to have a humane workplace, without really believing it himself, so it comes out a big mess.
Most of the book, though, is an interesting look inside two of the Aeron-infested workplaces of the fin-de-siècle, presenting a picture of young and diverse liberal arts majors jammed in a big room together playing loud music and patting themselves on the back for having such cool unjobs. Rather than focusing on the business stories behind Razorfish and 360hiphop.com, which we've presumably heard, this book tries to look at the work environment itself. Unfortunately it's written in a current-events American Studies tone rather than as an ethnography, which is too bad, because some anthropological skills would have made this a million times more interesting (and useful, to those of us that are trying to create humane workspaces while creating successful businesses).
This week, both InformationWeek and Baseline have feature stories about new technology at Delta Air Lines, so it's a good opportunity to benchmark Baseline, a new IT magazine from Ziff Davis, to its old school press-release republishing counterparts.
The difference: while the InformationWeek article is just a poor rewrite of bland warmed-over press releases and vague generalizations, the Baseline article is detailed, interesting, and specific. For example Baseline noticed that Delta's new information system crashed for two hours at the worst time possible: when I was in New Orleans trying to reschedule my flight home during last month's blizzard. InformationWeek didn't mention the outage. In fact the whole Baseline article looks like it was written by an investigative journalist. It's about time the industry press got its act together. Go for the free subscription, it's worth it.
Eric Sink analyzes all the leaky abstractions he just used to create Vault, SourceGear's new source control system, written in .NET. “I've used Java extensively, and I loved the productivity gains we got during the beginning and middle of the development cycle. But things got ugly at the end. All those layers of abstraction started contributing to our bug list. I've been involved in a couple of projects which completely failed because Java was chosen.... This success stands as a testimony to how incredible .NET really is.”
William Poundstone, who wrote an excellent book on game theory, has a new book about Microsoft interview questions, “How Would You Move Mount Fuji? Microsoft's Cult of the Puzzle”
1110 posts over 13 years. Everything I’ve ever published is right here.
There’s a software company in New York City dedicated to doing things the right way and proving that it can be done profitably and successfully.