search

svmiller / codename

Codename Generation for Organizations, People, Projects, and Whatever Else
http://svmiller.com/codename/
14 stars 4 forks source link

I need to improve `char2seed()` #1

Closed svmiller closed 3 years ago

svmiller commented 3 years ago

I'm astonished it happened this quickly, but I found two character seeds that produced identical numeric seeds. I won't spell both out here, but one of them is "A Reproducible Character Seed" (which is in the README). That's how I immediately knew it was a problem.

I'm labeling it a bug---it's not really a bug, per se---but it is a case where something is not working the way I want it to work. So, my intuition is this. I'm going to keep char2seed() as is for when version 0.2.0 goes on CRAN. For the next release after that, I'm going to mess with char2seed() to make it behave like this.

x <- "A Reproducible Character Seed" # set seed to
tmp <- c(1:26, 1:26, 0:9) # notice the change from 0:25 to 1:26. This is going somewhere.
names(tmp) <- c(LETTERS, letters, 0:9)

x <- gsub("[^0-9a-zA-Z]","",as.character(x))

xsplit <- tmp[ strsplit(x,'')[[1]] ]
xsplit

the_seed <- as.numeric(paste(xsplit, collapse="")) # We're just going to paste the split into a seed
the_seed # look at it

as.integer( the_seed %% (2^31-1) ) # same division as before

For old-timey's sake, I'll probably rename the current version to be something like char2seed_v1() but this will be the new char2seed() going forward.

svmiller commented 3 years ago

Fixed, going on CRAN soon hopefully.