Re: [lmi] C++11 transition step 0: replacing 0/NULL with nullptr

[Greg Chicares]

On 2016-05-17 21:56, Vadim Zeitlin wrote:
[...]
>  As I wrote there, most of the work was done by clang-tidy and I just
> reviewed its results and rebuilt (under all platforms, including using the
> official makefiles) and reran the tests (the rounding one indeed fails, as
> you recently mentioned, but no other ones seem to do, although I'm always
> nervous about this because the test suite doesn't clearly state how many
> tests have failed at the end, so it's easy to miss something, I'd like to
> change this...).

I'm reluctant to touch another low-level part of lmi. There's no urgent
need to change this. I do
  make $coefficiency unit_tests >../log 2>&1
  sed -f diagnostics.sed ../log 2>&1 |less -S
and then type
  /\*\*
in 'less'. I skip the six double asterisks that show problems in the
std::round() function that we don't use in production, and I skip the
intentional errors in the unit test of the unit-test system. Then
there's timer_test.

Fixing the actual problems demonstrated by failing tests is far more
valuable than summarizing them. I keep meaning to look up a message
you sent a while ago about 'timer_test'; I think you pointed me to
a wx unit test, but at this rate I won't have time to look at that
this year, so might I ask you to connect the dots for me and propose
a concrete change?

>  Needless to say, this PR affects a lot of the code, so it would be nice to
> merge it soon to avoid conflicts in the future

I already have a conflict in 'Makefile.am'. I can deal with that one
easily enough, but right now I really must merge some of my own
changes, as well as the new mortality-tables program (with some local
changes that I need to apply to it). And I have another conflict in
an old patch you sent a while back, which I'm trying to make time for.

Could we take a time-out to let me
 - finish my pending changes, and
 - expunge RCS '$Id$' globally, touching pretty much every file,
   as the first step toward migrating from svn to git
and then turn the PR spigot back on?