Closed andysturrock closed 8 years ago
OK fixed the bug now. There was a missing _nextByte = -1;
in TryReadMapLength
. All tests pass as expected.
I'll try putting the custom unpacker in the strict deserialiser next. Maybe then we can merge to master?
Apologies for the delay in responding to your issues/PRs. Have been catching up with family in Oz and my parent's internet was out for a few days. Back online now.
I had a few uninterrupted hours to work on this on the flight, somewhere over Iran I think. I found the issue I think you're referring to in this PR and fixed it in 31f4dd394ade088f6ecf8b003d1cd4ee6497c4fa. There's a solid test case there that stresses all combinations of values to draw out these kinds of issues.
I also switched to the custom unpacker in 09df0ff3d7bc9ec51373155c164e16070fd24c4d. This removes the dependency upon MsgPack.Cli in that project.
This PR was closed automatically when I deleted the previously merged custom-unpacker branch. From looking at it, I think all of this ended up on master one way or another.
Will respond to your other issues now.
The custom unpacker seems to fail when reading the map length and then a string. This first became apparent when I tried using the custom unpacker with the strict deserialiser. I'll take a look but I wondered whether you knew about this anyway. Or am I using it incorrectly?