FillerBase::commitRow bug shown in test_scantable.py

[Malte Marquarding]

Changeset r1867 introduces failures in test_scantable.py. The PKSFiller relies on the TableRow NOT being reinitialised.

I have tested introducing a new argument void commitRow(bool reset=false); to control if a row needs to be reinitialised and it works.

Waiting for Takeshi's comments. Please reassign to me after commenting in this ticket.

[Takeshi Nakazato]

I checked in a fix for the issue.

In the fixed version, commitRow() doesn't initialize TableRow. Instead, I modified setSpectra() to use define() method, not assignment (e.g. *specCol=spectrum), to set spectral values.

Could you test it?

[Takeshi Nakazato]

You don't need to modify commitRow(). I found a workaround to store spectra with different number of channel.

Please test current trunk.