Incorrect waveform length when using multiple invocations of "pigs wvag ..."

[phlipped]

When building a waveform using successive calls to "pigs wvag ...", the length of the waveform will sometimes be shorter than expected.

It appears that the total length of the wave form is (incorrectly) determined by using the delay of the final pulse in the waveform. If an earlier pulse has a very long delay, that delay will be ignored.

(underscore indicates delay)
Pulses A:  A1________A2_____________  # pigs wvag 1 0 10 0 1 15
Pulses B:  _____B1________B2__        # pigs wvag 0 0 5 2 0 10 0 2 4
----------------------------------------------
Expected:  A1___B1___A2___B2________
Actual:    A1___B1___A2___B2__

Example:

$ pigs wvclr
$ pigs wvsm 0
0
$ pigs wvag 1 0 10 0 1 15
2
$ pigs wvsm 0
25
$ pigs wvag 0 0 5 2 0 10 0 2 4
4
$ pigs wvsm 0
19

The issue is fairly easy worked around by making sure all pulse sets have the same total length, but nonetheless looks like a bug to me.

This also occurs when using pigpiod socket interface. I haven't tested with the C library directly.