If your ESP8266 behaves erratically, the cause could be in the power supply.
Not every experimenter is properly equipped with a reliable 3.3 V source, neither was I. So I built a downconverter to power the ESP8266 WiFi board. The LF33CV LDO regulator should supply 500 mA continuous: plenty of headroom for the 200-250 mA peak requirement.
I configured my modules to join the home WiFi network a startup. A simple test for their presence is to run a network ping (the smartphone and Fing app comes very handy). When powered with the USB-to-serial converter, the round-trip-time smartphone<>ESP8266 was of 10 ms or so. That is: send a small packet of data to the remote device and measure how long it takes to come back the acknowledgment.
When I moved to the stand-alone circuit I started getting lost packets exceptions and much higher RTT.
I looked at the 3.3 V line and noticed anomalous spikes when the LDO was used. Instead the DC line was much smoother with another 3.3 V source.
A simple extra 220 uF across 3.3 V and GND fixed it. Actually I even improved RTT of a couple of milliseconds.
So, decouple that 3.3 V line!