[lmi] Big census [Was: PR: Check that the run order is the same for all cells and the case]

[Greg Chicares]

On 2015-10-02 21:37, Vadim Zeitlin wrote:
[...]
>  On an unrelated note, while benchmarking the effects of this change with
> the "big census", I get this after ~40 (?) first cells
> 
>       While processing a decrease in policy month 0, policy year 2, 24 months 
> after beginning of most recent seven-pay period, cumulative premium should be 
> 24201.1 but is 40000; discrepancy is 15798.9.
>       [file /opt/lmi/src/git/ihs_irc7702a.cpp, line 1061]
> 
> It doesn't really bother me as the code being benchmarked is already
> executed by the time the error happens, but if you know immediately why
> does this error happen and, more importantly, how to fix it, I'd like to
> know about it because I thought that it would be nice to test the PDF
> output with 4200+ rows in this census to see if there are no unforeseen
> problems with that many pages.

I can reproduce this, but I can't guess how I made the mistake in
the first place. I'll send a replacement in personal mail.