Re: Bug in bttext_abbrev_convert()

On Tue, Jun 30, 2015 at 10:25 PM, Peter Geoghegan <pg(at)heroku(dot)com> wrote:
> It would be nice to always have a html report from gcov always
> available on the internet. That would be something useful to automate,
> IMV. Watching that regress over time might provide useful insight, but
> I only use gcov a couple of times a year, so that's not going to
> happen on its own.

I generated such a report just now, and noticed this:

1137 : tuplesort_performsort(btspool->sortstate);
214 1131 : if (btspool2)
215 0 : tuplesort_performsort(btspool2->sortstate);
216 :
217 1131 : wstate.heap = btspool->heap;
218 1131 : wstate.index = btspool->index;

The regression tests have zero coverage for this
tuplesort_performsort() "btspool2" case. That's a fairly common case
to have no coverage for, and that took me all of 5 minutes to find.
--
Peter Geoghegan