4 ms·
> The Churn is spending a week just to get a project you wrote a year ago to even run. How much effort would it be worth avoiding to spend a week every year ?
by hrktb 7y ago
> The Churn is spending a week just to get a project you wrote a year ago to even run.
How much effort would it be worth avoiding to spend a week every year ?
Sure the weeks can add up, but so does the time spent on low level abstraction, and refusing to adopt better tools when the whole environment in changing around (e.g. there is no mention of native mobile environments. Would it be churn use to Swift instead of non-ARC ObjectiveC ?)
- FigBug 7y agoI find Swift very bad for churn. Swift has had 5 versions in 5 years, with breaking api changes each time. Every answer on Stackoverflow about Swift has several answers, one for each api version. Any time you grab some Swift code from the web or an older project, it's not going to work. Avoiding the churn isn't an option since new Xcode versions drop support for old Swift versions. And only the two latest Xcodes will run on latest macOS. They even drop the support for the conversion tools. So if I go back to an old Swift project now, it won't compile in my Xcode, nor will my Xcode help convert the code to modern. My only option is to run an order version of Xcode in a VM to convert the code. If I'm writing a library I want other people to use or share between projects, I'll still do it in Objective-C. Apps I do in Swift but I find it annoying. I still have 15 year old non-ARC Objective C libraries. Why spend the time updating them when the are debugged and work fine? Every time I have to do a Swift version update I introduce bugs.
- hrktb 7y agoThat’s part of my point. Cutting yourself from new frameworks and hardware features just to cut churn would be a horrible tradeoff in most cases. There are some niches where churn can be mostly avoided, but I think churn is usually a fact of life we could just embrace at a healthy pace. From the opposite angle, a field with extremely low churn would seem suspicious to me. For instance I would expect any language with no significant update in the last 10 years to have abysmal unicode support.