[Libreoffice-commits] core.git: sal/rtl

[Mike Kaganski (via logerrit)]

 sal/rtl/math.cxx |   20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

New commits:
commit cb56edb177f4db5b9cc4d140543c4b11d41ef1b0
Author:     Mike Kaganski <mike.kagan...@collabora.com>
AuthorDate: Fri Feb 18 01:43:32 2022 +0300
Commit:     Mike Kaganski <mike.kagan...@collabora.com>
CommitDate: Fri Feb 18 05:04:39 2022 +0100

    Define policy to optimize to_decimal to our use case
    
    We don't need it to strip trailing zeroes (we do that ourselves anyway);
    also we don't need it to handle negatives.
    
    Makes doubleToString ~5% faster in my testing.
    
    Change-Id: Ie3c4e3fec1899364af54fda3f7141678a95359d0
    Reviewed-on: https://gerrit.libreoffice.org/c/core/+/130120
    Tested-by: Jenkins
    Reviewed-by: Mike Kaganski <mike.kagan...@collabora.com>

diff --git a/sal/rtl/math.cxx b/sal/rtl/math.cxx
index edb894cb7820..b9a0a3348757 100644
--- a/sal/rtl/math.cxx
+++ b/sal/rtl/math.cxx
@@ -300,6 +300,9 @@ void doubleToString(typename T::String ** pResult,
         return;
     }
 
+    // sign adjustment, instead of testing for fValue<0.0 this will also fetch 
-0.0
+    bool bSign = std::signbit(fValue);
+
     if (std::isinf(fValue))
     {
         // #i112652# XMLSchema-2
@@ -311,7 +314,7 @@ void doubleToString(typename T::String ** pResult,
             nResultOffset = 0;
         }
 
-        if (std::signbit(fValue))
+        if (bSign)
             T::appendAscii(pResult, pResultCapacity, &nResultOffset,
                            RTL_CONSTASCII_STRINGPARAM("-"));
 
@@ -321,17 +324,18 @@ void doubleToString(typename T::String ** pResult,
         return;
     }
 
-    decltype(jkj::dragonbox::to_decimal(fValue)) aParts{};
+    if (bSign)
+        fValue = -fValue;
+
+    decltype(jkj::dragonbox::to_decimal(fValue, 
jkj::dragonbox::policy::sign::ignore,
+                                        
jkj::dragonbox::policy::trailing_zero::ignore)) aParts{};
     if (fValue) // to_decimal is documented to only handle non-zero finite 
numbers
-        aParts = jkj::dragonbox::to_decimal(fValue);
-    else
-        aParts.is_negative = std::signbit(fValue); // Handle -0.0
+        aParts = jkj::dragonbox::to_decimal(fValue, 
jkj::dragonbox::policy::sign::ignore,
+                                            
jkj::dragonbox::policy::trailing_zero::ignore);
 
     int nOrigDigits = decimalDigits(aParts.significand);
     int nExp = nOrigDigits + aParts.exponent - 1;
     int nRoundDigits = 15;
-    if (aParts.is_negative)
-        fValue = -fValue;
 
     // Unfortunately the old rounding below writes 1.79769313486232e+308 for
     // DBL_MAX and 4 subsequent nextafter(...,0).
@@ -461,7 +465,7 @@ void doubleToString(typename T::String ** pResult,
     assert(nBuf <= 1024);
     typename T::Char* pBuf = static_cast<typename T::Char*>(alloca(nBuf * 
sizeof(typename T::Char)));
     typename T::Char * p = pBuf;
-    if (aParts.is_negative)
+    if (bSign)
         *p++ = '-';
 
     bool bHasDec = false;