Some time back I did a one-shot (or pulse extender) board that showed me a problem with hand soldering ceramic capacitors. Now that I have figured out how to do loop powered devices it seemed like a fun circuit to re-implemented as a loop powered device. I reflowed it in a toaster oven and there is no sign of thermal shock to the timing capacitor (on the previous board I hand soldered the ceramic timing cap, and the one-shot time kept reducing). This circuit seems more valuable after seeing that some flow meters use a reed and a spinning magnet to magnetize the reed and make it close (a flow pulse with lots of bounce or glitches). For such a flow meter I would want to use a small timing capacitor to extend the initial reed closure. When the reed bounces the barrage of capture events is difficult to deal with.
http://epccs.org/indexes/Board/OneShot/
My One-Shot board got an update and now seems to be working the way I want.
The power up is predictable (on ^2), after adding a capacitor and resistor that holds the reset low while power is applied.
I plan to use it with flow meters that have reed switches and other magnetic reed switch projects. Basically is is for magnetic reed switch sensors on things when I don't want to deal with a barrage of the bounces but do want the earliest possible indication. The bounces are simply masked during the one-shot time.
Water Proofing idea
The only problem with a one-shot approach is when you get noise in your pickup circuit - then you get false triggers because it triggered on a narrow spike (and it is not unusual that where you find flow meters, there are lots of other noise producing things like pumps etc.) My preferred approach to something like that (same thing I used for the cub scout derby timer 
) is to sample every ms or so. If I get 3 hits in a row, then it is valid, not noise (or somebody's flash in my case) - then for better timing, I back out the 3ms it took me to get the "valid" detection. These days with the cost of little single chip micros, it is easy to implement it on a small board like you have including signal conditioning (spike catching etc.). Just my $0.02
Precise pulse timing is important with flow meters, that is how flow rate gets determined. Validating the pulse with a debounce routine to make sure it is not noise is an option, but I want to use the capture hardware (e.g. Uno's ICP1 on pin 8 ) to measure the true time of the flow pulse. If a noise spike can cause the trigger input to falsely drop bellow its threshold on this circuit then that will be impressive. If the electronics connected is still working and I get a timer capture that will be even more impressive. I want to retain the earliest indication of a flow pulse for measurement quality, and mask the bouncy stuff so the MCU does not try to capture it. It transmits over a twisted pair (CAT5 shown) which I've tested on a 568-meter line. I have had very good luck with a twisted pair in noisy environments in the past. I guess if there is an occasional noise spike I will look for anomalous flow rates (or times) and toss the false flow pulses out of the data set.
I would like a programmable masking time (one-shot or pulse extender), but I'm not ready to try a loop powered MCU project yet. I also need to consider a way to program the time from the data acquisition side.