Ask YC: Coping with frustration a la setup/dependency/documentation hell?

nostrademons · on May 16, 2008

Kinda confused...is the question about the specific problem of learning new libraries, the general problem of setup/dependency/documentation hell, or the even more general problem of dealing with frustration?

For libraries, I just accept that it'll take a long time to get familiar with them, and dive in. That's what Hello World is for, anyway. If I've got them setup and working and still have a tough time mastering the library (bad documentation, for example), I'll often just ditch the library.

For setup/dependency/documentation hell, depends on the language. I use setuptools for all my Python-related code, and use it's dependency-management functionality whenever libraries support it. Small 3rd-party libraries (no more than a source file) get pulled directly into our source code repository, larger ones are easy_installed, with a shell script to take care of everything that isn't available via easy_install. HTML, CSS, and images get packaged along with the web framework code they're associated with, and installed via setuptools. I wrote a custom documentation/dependency analyzer for JavaScript, and use that for all our JavaScript. It's wrapped in a setuptools package that just consists of a library of build scripts, the JavaScript being pulled in through pkg_resources.

As for frustration, I usually go do something else. Oftentimes, when I return to the difficult part, I find it's become irrelevant.

whacked_new · on May 16, 2008

Thanks for the reply, and sorry about the confusion. This is a question about frustration in general; I thought the most salient example for fellow readers would be the hair-pulling craziness from dependency hell and that sort.

Taking the mind off the problem is one way, but subconsciously you know the problem is still there; so for example, if I decide to go treat myself a snack, I wouldn't enjoy the taste as much because the stress is still there, and the "time elapsed since you started" keeps increasing, which is annoying to think about when you ask yourself "how much did I accomplish in the last X hours".

Say you are 2 hours into the onset of some extreme frustration. In your experience, do the benefits of leaving for 20 minutes and coming back usually outweigh the costs of decreased efficiency from working in a bad mood?

nostrademons · on May 16, 2008

I actually meant a longer timescale...like leaving the problem alone for a week while I work on some other aspect of the product, and then coming back to it. Like if I'm stuck on a UI library, I'll go work on backend stuff instead. About half the time, the solution to my original problem is obvious when I return to it, and another quarter of the time, the problem is irrelevant and I need to rewrite it anyway.

If I'm stuck on everything, I'll work on a side project. That's how Scrutiny, Bootstrapacitor, ArcLite, and Randomicity were done. (Most of the time, the side project itself amounts to nothing, but it serves its purpose of getting me unblocked.)

This is one reason why I find projects often accelerate as I get farther along in them: they generate other productive tasks that I can do when I get stuck. I'm less likely to work on something completely unrelated then, so overall project velocity increases.