[R] Strange behavior with merging algorithm

[Gene]

I wrote some code to merge some columns from different data frames  
into a new data frame based on mapping data in one of the columns.  I  
know that I can use the merge function for this, and now, in  
retrospect, I don't know why I didn't.  However, it seems like the  
code that I wrote should work but didn't.  I found a work-around  
which I demonstrate here.  Can anyone explain why the first section  
doesn't work but the second one does?  And, for tuning my basic R  
skills, are there better ways for doing this manually (i.e. not using  
'merge') than how I do it?

Thanks.

# final resting place for data
complete.data <- data.frame(positions=good.positions)
j <- 1

# generate mapping information
map <- match( good.positions, orig.data$positions)
is.pos.present <- ! is.na(map)
good.pos <- map[ is.pos.present ]


## this doesn't seem to work
complete.data[is.pos.present,j] <- orig.data$values[good.pos]


## this does work
tmp <- rep(NA,length(is.pos.present))
tmp[is.pos.present] <- orig.data$values[good.pos]
complete.data[,j] <- z