vint64 removal, and my work plans.

Eric S. Raymond esr at thyrsus.com
Tue Dec 1 23:51:54 UTC 2015


In preparing to remove vint64s I found a note I'd left myself in a comment
and forgotten.  The ISC library requires the C99 int64_t/uint64_t types to
exist.  Therefore we won't lose any portability by removing vint64s.

But I'm not going to do it yet, because several of the modules that would
be modified have unit tests.  Insteads I'll wait until Amar finishes those
and I can use them to test for regressions as I take out the vint64 code.

This removal will only be about 1 KLOC, but it's a *nasty* 1 KLOC - hard
to understand, and good riddance when it's gone.

My current main goal is still TESTFRAME.  I think I now know that I don't need
to do anything special about getaddrinfo calls, because the intercept code will
record and play back numeric IP addresses.  I also think I know where to record
packet receives.

However, I've also figured out that replay mode will involve conditioning
out a lot of complex network plumbing - basically, all the logic for repeatedly
polling the UDP sockets for current associations.

I don't more than vaguely and generally understand that code yet.  I think
the time has come for me to at least start writing a tour document - analyze
what's there and describe how it works and what the main invariants are.
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>


More information about the devel mailing list