fwrite(data.table(1), dec=',')
# Error in fwrite(data.table(1), dec = ",") : dec != sep is not TRUE

But this doesn't matter on a 1-column file where sep is not used.

As an aside (since we can already create these ambiguous files today):

It does create the potential for confusion where this file could either be 1- or 2-columns for fread:

fwrite(data.table(c(0.1, 0.2)), dec=',', sep='\t')
# V1
# 0,1
# 0,2

Basically you can't rely on the sep= auto-detection for such files any more -- even if there are 100s of x,y fields and only one header column, fread() doesn't guess that "maybe dec=',' and there's only one column".