Right, here's the parts that required less trivial fixes which couldn't be done whilst ill. There's a few points to be aware of buried in here:
I've changed gadgetstock to only treat initialconditions as a bare component (i.e. it's component-like but has no brackets around the name). Before it would treat any line without a \t as a component, which given parsing has changed elsewhere to not leave trailing \t on missing values, was causing a host of file-reading bugs, as the stockfile was ~arbitrarily broken up into components. This is good, but you might have code that relies on previous behaviour.
Postambles after ; -- data -- components was returning one string, not a list of lines with commas removed as elsewhere. I've fixed it to be consistent, but again you might have been relying on what happened before.
I put the df_component machinations back again so time variable files can get their table read as a data.frame, not a list. But because of the changes with table parsing, multiple-whitespace-separated columns aren't supported any more. This may trip up other things than the unittests.
I've restored the initialcondition generation from the numberfile, sounds like it leans towards being the right thing to do, even if not heavily used.
Right, here's the parts that required less trivial fixes which couldn't be done whilst ill. There's a few points to be aware of buried in here:
initialconditions
as a bare component (i.e. it's component-like but has no brackets around the name). Before it would treat any line without a\t
as a component, which given parsing has changed elsewhere to not leave trailing\t
on missing values, was causing a host of file-reading bugs, as the stockfile was ~arbitrarily broken up into components. This is good, but you might have code that relies on previous behaviour.; -- data --
components was returning one string, not a list of lines with commas removed as elsewhere. I've fixed it to be consistent, but again you might have been relying on what happened before.df_component
machinations back again so time variable files can get their table read as a data.frame, not a list. But because of the changes with table parsing, multiple-whitespace-separated columns aren't supported any more. This may trip up other things than the unittests.