Fast Diff

[brodieG]

Provide a more useful version of ses that has minimal overhead. Initial implementation could be some way of setting all flags to the minimal effort level on diffChr.

[brodieG]

Merging #84 into this issue. Note that in:

x <- y <- 1:1e6
x[sample(1e6, 100)] <- sample(1e6, 100)
library(diffobj)
system.time(diffChr(x, y))
#    user  system elapsed
#  47.582   0.267  47.829
system.time(diffChr(x, y, word.diff=FALSE))
#    user  system elapsed
#  13.566   0.344  13.902
system.time(diffChr(x, y, word.diff=FALSE, ignore.white.space=TRUE))
#    user  system elapsed
#  13.203   0.585  13.779
system.time(diffChr(x, y, word.diff=FALSE, ignore.white.space=TRUE, convert.hz.white.space=FALSE))
#     user  system elapsed
#    8.711   0.104   8.811

word.diff and convert are the big issues. Also, still much slower than ses

system.time(ses(x, y))                                                                   user  system elapsed
#   0.829   0.016   0.844

[brodieG]

Actual implementation will be a no frills version of diffChr close to what tools::Rdiff does.

[brodieG]

No demand for this.