Last week (and the week before) we heard some compelling arguments for the support of legacy platforms. We also learned of the customer service advantages this kind of support can lend a company. But what if there is no company? What if, instead of a team of coders on a project, you find yourself the sole developer? And what if that project, instead of being a simple arcade game from the ’90s, is nothing short of an award-winning replacement for the Finder itself?
This is the situation Steve Gehrman, sole developer of the acclaimed utility Path Finder, found himself in when contemplating the fourth major release of his über file manager. In moving forward, he decided to eschew legacy platforms and, with the help of tech support and web design whiz Neil Lee, make PF4 a Tiger-only app. How has this strategy worked out for these two gentlemen of Cocoatech? “It was the right decision,” says Steve.
“Lots of developers try to be the nice guy and support older versions of the OS,” he says. “It’s a great ideal, but not worth the effort. In some apps it might be simple to support both [old and new platforms], but when you have a large, complex, UI heavy app, supporting multiple versions of an OS is always much more difficult than it appears.”
Difficult how? “When a new OS is released, there are tons of bug fixes and new technologies,” Steve explains. “Many times, code that worked on one version of the OS doesn’t work on the other. Usually what happens is I get the new version of the OS and start fixing bugs. I think everything is great until I try to test it on the old version of the OS. Now running on the old OS is broken.”
So choosing to support only the latest OS is more about stability than new features like Core Data and Spotlight?
“Technologies like Spotlight are something I wanted to use,” grants Steve, “but the bottom line is that I would rather ship a rock-solid, full-featured, 10.4-only app that has been fully tested than a mediocre app that supports an older OS that is obsolete.”
Neil Lee agrees. “The #1 motivation [in going Tiger-only] was to take advantage of the new stable frameworks and bug fixes to Cocoa,” he says. “Apple provides developers with a ton of cool toys, but bugs and weird behavior are part of any complex system. Previous versions of Path Finder had a lot of workarounds and funky moves to handle stuff like that, and moving 10.4-only meant that Steve could rip out a lot of this and clean things up.”
Still, it would be possible to offer a separate version based on old code that incorporated some of PF4’s new features. True, this would be built on a more buggy foundation, but people could update to Tiger if they wanted the stability. Would offering customers a choice like that be beneficial?
Steve is adamant. “I never considered it.”
“I could try to have two sets of code,” he continues, “but then you waste too much time debugging and testing, and your code becomes too ugly and hard to maintain.”
“It comes down to resources,” Neil puts in. “It’s just Steve doing all of the programming and me handling the lion’s share of the tech support, and having to fork our efforts into maintaining two separate codebases and two separate OS’s bugs and quirks would have really affected our ability to deliver a good product.
“To paraphrase Homer: we didn’t want to do Path Finder 4 half-assed; we wanted to put our entire ass into it!”
Once the decision is made to cut support for older systems, the question quickly becomes, “Will our customers upgrade to stay with us?” Path Finder has very loyal users, but what percentage are now running Tiger?
“We don’t track our customer base’s system configuration, but a total ballpark guess based on sales would be at least 70% and maybe even as high as 80%,” estimates Neil. “I would assume that people who are interested in using a third-party file browser would tend to be be more technically savvy and running the most recent version of the OS.”
“I always update to the latest OS,” adds Steve. “Most paying customers will also pay for the new OS. Each version is so much better than the last that it doesn’t make sense to stay on the old one. Apple fixes tons of bugs with each release, so why continue using a buggy version of the OS? If automobiles came with $100 upgrades everyone would update. I don’t know why an OS is any different. I spend more time in front of my computer than in my car.”
I can sympathize with that! But speaking of updates, OS X 10.5 will be previewed at WWDC in August. Can we expect a Path Finder 5?
“I’m sure 10.5 will be so much better than 10.4 that I will have no choice,” says Steve. “With every release, Cocoa and the other APIs are so much better. The programming APIs only change on major releases of the OS, so it’s basically a year’s worth of great stuff that I can now use in my app.”
Great stuff like what?
“Hopefully Cocoa will be improved,” Steve opines. “I’ve emailed a bunch of complaints and bug reports to Apple over the years and I think big changes are coming. The transition to OS X was such a huge task that Cocoa was somewhat neglected in my opinion. Hopefully I won’t be disappointed.
“Another thing I’m hoping for is better looking UI widgets and more standard UI animations. The most highly acclaimed app out there, Delicious Library, is all custom photoshopped buttons. I don’t have photoshop skills, so my app looks ugly in comparison. I want my app to look great without having to resort to hiring a full time photoshop guy to rework my UI.”
So with luck, it won’t be too long before we see an even more stable and flashy looking Path Finder 5. And we can assume it will be 10.5-only?
“Yes,” says Steve. “I have to compete against the 10.5 Finder, which is obviously 10.5-only, too.”