Actually, nevermind, I thought of a better way to achieve my ends that does not require this after all, although it might still be an interesting academic exercise ... -- bc
On Thu, Jul 24, 2014 at 9:06 PM, Bob Cronin <[email protected]> wrote: > I have an application in which my input takes the form of atoms separated > by slashes. For example: > > a/b/c or d/e/f/g or even h/i/j/k/l > > That is, the number of atoms delimited by slashes varies. > > From each input I need to generate "n" variants, where "n" equals the > number of slashes. > > For example, for a/b/c, I need to generate a/V/b/c and a/b/V/c. > > For d/e/f/g I need d/V/e/f/g, d/e/V/f/g and d/e/f/V/g. > > For h/i/j/k/l I need h/V/i/j/k/l, h/i/V/j/k/l, h/i/j/V/k/l and h/i/j/k/V/l. > > and so on ... > > More generally I suppose I can characterize the problem as the need to > output a variant consisting of the original input with a constant-value > field inserted after every field in the input except the last. > > Thoughts on how best to do this? > > This will need to be inserted into an existing pipeline that has some > pretty hairy things going on already and so it will be important to avoid > things that might end up causing stalls. Off the top of my head it doesn't > seem as though a solution to this would introduce any delays, but I admit > to being a bit fuzzy on that even after all these years. > -- > bc > > >
