Bugfix/2017 04 bgq fixes (#152)

* fixed typo with endianness test
* fixed type issues that appeared on bgq related to chars
* fixed typo in uberenv variant logic
* used signed char for bb64 decode table

on bgq, "char" defaults to an unsigned type with
gcc and xlc.

libb64 uses negative entries in its decoding table,
and the negative literals were being (silently) converted
to unsigned types, which broke the decoding logic.

* added more checks to bb64 smoke

Thanks to Kenny Weiss for the help tracking down
 and resolving these issues!

scripts/uberenv/packages/uberenv-conduit/package.py

@@ -221,7 +221,7 @@ def install(self, spec, prefix):
                 cfg.write("# sphinx from uberenv\n")
                 cfg.write("#" + cmake_cache_entry("SPHINX_EXECUTABLE",py3_sphinx_build_exe))
 
-        if not "+python" in spec or "+pyhton3" in spec:
+        if not "+python" in spec and "+python3" in spec:
             cfg.write(cmake_cache_entry("ENABLE_PYTHON","OFF"))
 
         #######################

src/tests/conduit/c/t_c_conduit_node_set.cpp

@@ -473,7 +473,7 @@ TEST(c_conduit_node_set, set_native_int_ptr)
         // set(...) semantics imply a copy -- mem addys should differ
         EXPECT_NE(&icav_ptr[i],&icav[i]);
     }
-    EXPECT_EQ(icav_ptr[5],-64);
+    EXPECT_EQ(icav_ptr[5],char(-64));
 
     // using short* interface
     conduit_node_set_short_ptr(n,isav,6);

src/tests/conduit/t_conduit_endianness.cpp

@@ -418,7 +418,7 @@ TEST(conduit_endianness, node_swap_using_explicit_funcs)
                                  8, // elem_bytes
                                  Endianness::LITTLE_ID); // set not match
 
-        n["test32"].endian_swap_to_machine_default();
+        n["test64"].endian_swap_to_machine_default();
         EXPECT_EQ(0x0102030405060708,test64.vuint64);
 
         // swap all back back so we can do a full test on n

src/tests/conduit/t_conduit_node_set.cpp

@@ -2294,7 +2294,7 @@ TEST(conduit_node_set, set_cstyle_int_array)
         // set(...) semantics imply a copy -- mem addys should differ
         EXPECT_NE(&char_ptr[i],&char_av[i]);
     }
-    EXPECT_EQ(char_ptr[5],-64);
+    EXPECT_EQ(char_ptr[5],char(-64));
 
     // short 
     n.set(short_av_a);

src/tests/thirdparty/t_libb64_smoke.cpp

@@ -54,9 +54,11 @@
 #include "b64/encode.h"
 #include "b64/decode.h"
 
-TEST(libb64_smoke, basic_use )
+
+std::string
+encode_decode(const std::string &val)
 {
-    std::string sin("test");
+    std::string sin(val);
 
     std::istringstream iss(sin);
     std::ostringstream oss;
@@ -77,6 +79,28 @@ TEST(libb64_smoke, basic_use )
 
     d.decode(iss,oss);
 
-    EXPECT_EQ(oss.str(),sin);
+    return oss.str();
+}
+
+
+TEST(libb64_smoke, basic_use )
+{
+    std::string t = "test";
+    EXPECT_EQ(encode_decode(t),t);
+
+    // test a shorter string ...
+    t = "t";
+    EXPECT_EQ(encode_decode(t),t);
+
+    // test an empty string
+    t = "";
+    EXPECT_EQ(encode_decode(t),t);
+
+    t = "test a longer string";
+    EXPECT_EQ(encode_decode(t),t);
+
+    t = "test an even longer string with more vowels.";
+    EXPECT_EQ(encode_decode(t),t);
+
 }
 

src/thirdparty_builtin/libb64-1.2.1/src/cdecode.c

@@ -15,10 +15,11 @@ MOD FOR CONDUIT: This warning suppression deals with issue w/ gcc on BG/Q.
 
 
 #include <b64/cdecode.h>
+#include <stdio.h>
 
 int base64_decode_value(char value_in)
 {
-	static const char decoding[] = {62,-1,-1,-1,63,52,53,54,55,56,57,58,59,60,61,-1,-1,-1,-2,-1,-1,-1,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,-1,-1,-1,-1,-1,-1,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51};
+	static const signed char decoding[] = {62,-1,-1,-1,63,52,53,54,55,56,57,58,59,60,61,-1,-1,-1,-2,-1,-1,-1,0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,-1,-1,-1,-1,-1,-1,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51};
 	static const char decoding_size = sizeof(decoding);
 	value_in -= 43;
 	if (value_in < 0 || value_in >= decoding_size) return -1;
@@ -35,7 +36,7 @@ int base64_decode_block(const char* code_in, const int length_in, char* plaintex
 {
 	const char* codechar = code_in;
 	char* plainchar = plaintext_out;
-	char fragment;
+	signed char fragment;
 
 	*plainchar = state_in->plainchar;
 
@@ -89,7 +90,7 @@ int base64_decode_block(const char* code_in, const int length_in, char* plaintex
 				{
 					state_in->step = step_d;
 					state_in->plainchar = 0;
-					return(int)(plainchar - plaintext_out);
+					return (int)(plainchar - plaintext_out);
 				}
 				fragment = (char)base64_decode_value(*codechar++);
 			} while (fragment < 0);