livie
/
phoenix-firestorm
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Suppress non-useful NaN test in llsd_new_tut.cpp

master
Nat Goodspeed 2014-10-24 11:55:37 -07:00
parent 2ed557e16b
commit 80fdd85a0c
1 changed files with 12 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
indra/test/llsd_new_tut.cpp
View File

@ -277,8 +277,18 @@ namespace tut
v = 0.5; checkConversions("point5", v, true, 0, 0.5, "0.5");
v = 0.9; checkConversions("point9", v, true, 0, 0.9, "0.9");
v = -3.9; checkConversions("neg3dot9", v, true, -3, -3.9, "-3.9");
v = sqrt(-1.0); checkConversions("NaN", v, false, 0, sqrt(-1.0), "nan");
// Get rid of NaN test. First, some libraries don't reliably return
// NaN for sqrt(-1.0) -- meaning that I don't even know how to
// portably, reliably produce a NaN value. Second, we observe failures
// on different platforms in the asString() test. But LLSD's
// ImplReal::asString() does not itself recognize NaN! It merely
// passes the value through to llformat(), which passes it through to
// the library vsnprintf(). That is, even when we do produce NaN,
// we're not testing any LLSD code: we're testing the local library's
// vsnprintf() function, which (empirically) produces idiosyncratic
// results. This is just not a good test case.
// v = sqrt(-1.0); checkConversions("NaN", v, false, 0, sqrt(-1.0), "nan");
v = ""; checkConversions("empty", v, false, 0, 0.0, "");
v = "0"; checkConversions("digit0", v, true, 0, 0.0, "0");
v = "10"; checkConversions("digit10", v, true, 10, 10.0, "10");