[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[lmi-commits] [lmi] master ba0aad9 2/4: Improve documentation
From: |
Greg Chicares |
Subject: |
[lmi-commits] [lmi] master ba0aad9 2/4: Improve documentation |
Date: |
Thu, 11 May 2017 16:53:07 -0400 (EDT) |
branch: master
commit ba0aad9704d4629c48cde238dbbf298d15f87c74
Author: Gregory W. Chicares <address@hidden>
Commit: Gregory W. Chicares <address@hidden>
Improve documentation
---
expression_template_0_test.cpp | 12 ++++++++----
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git a/expression_template_0_test.cpp b/expression_template_0_test.cpp
index 64639e7..c6cb485 100644
--- a/expression_template_0_test.cpp
+++ b/expression_template_0_test.cpp
@@ -79,10 +79,9 @@ struct greater_of
}
};
- // Global variables for timing tests. It would be in better taste
- // to pass them as arguments, using std::bind. However, that
- // would rule out using some compilers (e.g., borland), and it's
- // best to test this with as many different toolsets as possible.
+ // Global variables for timing tests. They could alternatively be
+ // passed as arguments, e.g., by using std::bind, but that would
+ // increase complexity in return for no real benefit.
int g_length = 1;
@@ -309,6 +308,11 @@ void mete_valarray_typical()
}
#if defined USE_UBLAS
+// This is never actually called. It is incomplete: it is not known
+// whether the operations measured by other /mete.*typical/ functions
+// can even be expressed with this library; and that doesn't matter,
+// because other tests show it to be slower than its competitors.
+//
// This library seems to provide most of what we need as named
// functions like prod(), not as the overloaded operators that
// we'd prefer for clarity.