Quoting Brent Yorgey
On Thu, Jun 18, 2009 at 08:30:20AM -0400, wagnerdm@seas.upenn.edu wrote:
This is probably something of a controversial patch: it adds two dependencies. I don't think the "parsec" dependency is a big deal, since
Hmm. I'm not sure how I feel about the extra dependencies. (Literally---I'm not using "I'm not sure how I feel" as a euphemism for "I don't like it"). But I thought I should point out
Right, the more I think about it, the more I recognize that extra dependencies of any kind could be a problem for xmonad-contrib. I recognize that the target audience includes non-Haskell people who can't really be expected to (and shouldn't really have to) understand the Haskell library landscape. On the other hand, the library landscape is there precisely so that people who do know Haskell can use it. =) I haven't decided what I want to do about this yet. The way I see it, I have two choices if I want to publish this for other people to use (and in either case, the patch I sent previously should not be applied): 1. Rip out code from Data.List.Split and convert the Parsec stuff to ReadP. This sucks because I'd have to waste time rewriting the code, it would make me feel icky to duplicate code from other libraries, and it wouldn't really settle the issue for people who want to write contributions in the future that depend on libraries with less obvious counterparts. 2. Publish a separate package. This sucks because it means it wouldn't be in the "blessed" collection of code we call xmonad-contrib; people would be less likely to find it, they would waste more time duplicating it if it was something they wanted, and they would file fewer bug reports. I am open to suggestions. ~d