To increase the usefulness of the example, I added the measurement of
the complete cycle, which is the important figure. The reading time
itself, though interesting, is not the limiting factor according to the
datasheet.
Additionally, I updated the printing instructions to create a nicely
formatted output and limited the number cycles necessary to reach "del
== 0".
In multihreaded environment (I used NilRTOS) its usefull to disable interruppts while reading the sensors. However delay() requires interrupts to be enabled. If you read more than one sensor in a sequence the 2nd one gets spurious DHTLIB_ERROR_CONNECT errors. Timing issues with the wakeup delay().
They don't used to happen it interrupts are enabled. Replacing delay() with delayMicroseconds() solved this issue.
+ replaced micros() with adaptive loopcount
+ removed DHTLIB_INVALID_VALUE (user should decide if value is usable)
+ more detailed errors per phase of handshake
+ added DHTLIB_ERROR_CONNECT
+ added DHTLIB_ERROR_ACK_L DHTLIB_ERROR_ACK_H
+ added fix for pre 1.0 (not tested)
detection of the zero/one is doen by measuring the whole bit LOW+HIGH
+ improved wait for acknowledge
+ merged two loopCounters into one.
+ layout
+ dht22_test sketch collects statistics (every 20 reads)
