Extract lookup() to a small "tools" package

[krlmlr]

I just installed qdap to use its lookup function, and I was really stunned about the number of packages that are pulled in. I don't do NLP, so I'm not really using many of these packages.

There doesn't really seem to be a good alternative to lookup, cf. http://stackoverflow.com/questions/15303283/how-to-do-vlookup-and-fill-down-like-in-excel-in-r . Would you consider extracting this function (and its associated operators) to a small, dependency-free package? Other candidates would be hash, hms2sec, and generally anything else that is not specific to your domain.

[trinker]

This seems like a reasonable request. It would be after the end of semester. I think I'd figure out a script that builds everything by taking from qdap so there's less maintenance involved if these functions are updated.

May I ask if http://cran.r-project.org/web/packages/hash/hash.pdf would suit your needs?

[krlmlr]

Would you be offering these functions in two packages? Why not Import the "tools" package from qdap? (People will have to load the new package, but this seems to be a minor inconvenience.) Or, if you Depend on the "tools" package, the functions should be still available in the search path.

The hash package is an option, but I like the elegance of the %l% and %l+% operators. Also, from my understanding, lookup does not convert keys to character before performing the lookup, in contrast to hash.

[trinker]

The original audience for qdap are people in my field. These people have never touched the command line (generally) and I try to make using qdap as pain free as possible. That means everything that was needed was in one manual (help manual) and everything loaded by default. That's why qdap is a bloated beast of a package.

But I have done what you suggest with qdapDictionaries. Everything can be contained in the qdap vignettes. So this may not be a bad approach. I'd probably do a name like qdapTools which keeps the packages together name wise but the name is less informative for people looking for a lookup type of function.

[trinker]

167 is a connected issue in that I'll address both of these in one fell swoop. Projected time to work through these 2 issues is mid May.

[trinker]

I have begun separating these functions into a separate package: https://github.com/trinker/qdapTools

[trinker]

This has occurred though not pushed to CRAN yet. A bit oft testing and feedback should occur first.

I'm closing this and if we have more issues could we discuss them at the qdapTools issues page?

[trinker]

I pushed to CRAN today: http://cran.r-project.org/web/packages/qdapTools/index.html

Kirill your last name got jacked up because I tried to use non ascii characters. I changed to Muller but if you have a way to use the accent mark in the DESCRIPTION file please pull request.

The next step is to start removing these functions from qdap and using importFrom qdapTools where needed and add as a Dependency

[krlmlr]

Thanks a ton for your efforts and for your ultrafast response! Muller is ok, I really wonder if anyone still uses R on a platform that has no Unicode support...

[trinker]

Yeah I love using R but when I saw this post http://www.r-bloggers.com/a-weekend-with-julia-an-r-users-reflections/ on julia's use/handling of unicode characters I was definitely intrigued.

[krlmlr]

I've heard about that, though I think you can use Unicode identifiers in R (at least if your system is using a Unicode locale):

> ü <- 3
> ü
[1] 3

Anyway... Are you adding a release to https://github.com/trinker/qdapTools/releases ?

[krlmlr]

Wrong:

> € <- 4
Error: unexpected input in "�"

[trinker]

Thanks for the release suggestion. Didn't know you could do that.

[krlmlr]

You're welcome. I usually copy the relevant parts of NEWS.md into the description of the release.