Merge pull request #125 from UDST/develop

v0.4.4 release

CHANGELOG.md

@@ -1,3 +1,10 @@
+v0.4.4
+======
+
+2019/9/4
+
+* Restores support for pre-C++11 compilers.
+
 v0.4.3
 ======
 

README.md

@@ -1,4 +1,6 @@
-[![Build Status](https://travis-ci.org/UDST/pandana.svg?branch=master)](https://travis-ci.org/UDST/pandana)
+![Version](https://img.shields.io/badge/version-0.4.4-blue.svg)
+[![Travis build status](https://travis-ci.org/UDST/pandana.svg?branch=master)](https://travis-ci.org/UDST/pandana)
+[![Appveyor build status](https://ci.appveyor.com/api/projects/status/a90kvns7qe56kg57?svg=true)](https://ci.appveyor.com/project/smmaurer/pandana)
 [![Coverage Status](https://coveralls.io/repos/github/UDST/pandana/badge.svg?branch=master)](https://coveralls.io/github/UDST/pandana?branch=master)
 
 # Pandana

appveyor.yml

@@ -0,0 +1,24 @@
+build: false
+
+environment:
+  matrix:
+    - PYTHON: 2.7
+    - PYTHON: 3.6
+
+init:
+  - "ECHO %PYTHON%"
+
+# The goal here is mainly to confirm that Pandana compiles in Windows; we're not
+# running any of the CI tests
+
+# Cython and NumPy need to be present before building Pandana. Some other
+# runtime requirements aren't installing well with Pip in win-py27, so I'm just
+# putting all of them into the Conda environment
+
+install:
+  - "set PATH=C:\\Miniconda3;C:\\Miniconda3\\Scripts;%PATH%"
+  - conda config --append channels conda-forge
+  - "conda create --name test-env python=%PYTHON% pip cython matplotlib numpy osmnet pandas requests scikit-learn pytables --yes --quiet"
+  - activate test-env
+  - python setup.py install
+  - conda list

docs/conf.py

@@ -57,9 +57,9 @@
 # built documents.
 #
 # The short X.Y version.
-version = '0.4.3'
+version = '0.4.4'
 # The full version, including alpha/beta/rc tags.
-release = '0.4.3'
+release = '0.4.4'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.

docs/installation.rst

@@ -16,6 +16,8 @@ Pandana is hosted on Conda Forge::
     conda install pandana --channel conda-forge
 
 
+.. _pip:
+
 Pip (requires local compilation)
 --------------------------------
 
@@ -25,6 +27,10 @@ Pandana is also hosted on PyPI::
 
 Pandana's C/C++ extension will compile automatically if the right tools are present. See below for troubleshooting.
 
+Pip can have trouble installing Pandana's requirements in Windows Python 2.7; try Conda instead.
+
+
+.. _github:
 
 GitHub source code
 ------------------
@@ -35,6 +41,8 @@ If you'll be modifying the code, you can install Pandana from the `GitHub source
     cd pandana
     python setup.py develop
 
+You'll need to have Cython and NumPy installed before running :code:`setup.py`. 
+
 Pandana's C/C++ extension will compile automatically if the right tools are present. See below for troubleshooting.
 
 
@@ -43,16 +51,10 @@ Compiling locally
 
 Building Pandana from source requires C/C++ compilers. On Linux and Mac these are usually already present, but read on for more information.
 
-.. note::
-    Pandana's C/C++ code references some libraries from NumPy, so you'll need to have NumPy fully installed before running Pandana's setup script.
-
-.. note::
-    Pandana uses OpenMP to parallelize computations --- compiling without OpenMP support will still work but won't allow multi-threading. 
-
 Compiling Pandana in Linux
 ~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-Pandana's setup script expects a version of the GCC compiler with support for the C++11 standard and OpenMP. This appears to be GCC 4.8+, but we haven't done extensive testing. If you run into problems, try doing a fresh install of the core build tools::
+Pandana's setup script expects a version of the GCC compiler with support for OpenMP. This appears to be GCC 4.8+, but we haven't done extensive testing. If you run into problems, try doing a fresh install of the core build tools::
 
     sudo apt-get install --reinstall build-essential
 
@@ -66,16 +68,19 @@ Certain older machines may need the `Microsoft Visual C++ 2008 SP1 Redistributab
 Compiling Pandana in OS X
 ~~~~~~~~~~~~~~~~~~~~~~~~~
 
-The default OS X compilers don't support OpenMP multi-threading. Use these commands to confirm that Xcode Command Line Tools are present and to install some newer compilers from Anaconda::
+The default OS X compilers don't support OpenMP, which Pandana uses for multi-threading.
+
+To allow multi-threading, use these commands to confirm that Xcode Command Line Tools are present and to install some newer compilers from Anaconda::
 
     xcode-select --install
     conda install llvm-openmp clang
 
-After installing Pandana, running :code:`examples/simple_example.py` will display the number of threads that Pandana is using.  
+Then follow the :ref:`Pip <pip>` or :ref:`GitHub source code <github>` instructions above.
 
-.. note::
-    If you get a compilation error like ``'wchar.h' file not found``, you can resolve it in macOS 10.14 by installing some additional header files::
+If you get a compilation error like ``'wchar.h' file not found``, you can resolve it in macOS 10.14 by installing some additional header files::
 
-        open /Library/Developer/CommandLineTools/Packages/macOS_SDK_headers_for_macOS_10.14.pkg
+    open /Library/Developer/CommandLineTools/Packages/macOS_SDK_headers_for_macOS_10.14.pkg
+
+After installing Pandana, running :code:`examples/simple_example.py` will display the number of threads that Pandana is using.  
 
 .. _Anaconda: https://www.anaconda.com/distribution/

pandana/__init__.py

@@ -1,3 +1,3 @@
 from .network import Network
 
-version = __version__ = '0.4.3'
+version = __version__ = '0.4.4'

pandana/tests/test_cyaccess.py

@@ -34,9 +34,9 @@ def net(nodes_and_edges):
 
     net = cyaccess(
         nodes.index.values,
-        nodes.as_matrix(),
-        edges.as_matrix(),
-        edge_weights.transpose().as_matrix(),
+        nodes.values,
+        edges.values,
+        edge_weights.transpose().values,
         True
     )
 
@@ -109,9 +109,9 @@ def test_memory_leak(nodes_and_edges):
         print i
         net = cyaccess(
             nodes.index.values,
-            nodes.as_matrix(),
-            edges.as_matrix(),
-            edge_weights.transpose().as_matrix(),
+            nodes.values,
+            edges.values,
+            edge_weights.transpose().values,
             True
         )
 '''

setup.py

@@ -102,7 +102,7 @@ def run(self):
 ## Standard setup
 ###############################################
 
-version = '0.4.3'
+version = '0.4.4'
 
 packages = find_packages(exclude=["*.tests", "*.tests.*", "tests.*", "tests"])
 

src/accessibility.cpp

@@ -1,7 +1,6 @@
 #include "accessibility.h"
 #include <algorithm>
 #include <cmath>
-#include <functional>
 #include <utility>
 #include "graphalg.h"
 
@@ -18,6 +17,23 @@ bool distance_node_pair_comparator(const distance_node_pair& l,
                                    const distance_node_pair& r)
     { return l.first < r.first; }
 
+
+double exp_decay(const double &distance, const float &radius, const float &var)
+{
+	return exp(-1*distance/radius) * var;
+}
+
+double linear_decay(const double &distance, const float &radius, const float &var)
+{
+	return (1.0-distance/radius) * var;
+}
+
+double flat_decay(const double &distance, const float &radius, const float &var)
+{
+	return var;
+}
+
+
 Accessibility::Accessibility(
         int numnodes,
         vector< vector<long>> edges,
@@ -350,17 +366,14 @@ Accessibility::aggregateAccessibilityVariable(
     double sum = 0.0;
     double sumsq = 0.0;
 
-    std::function<double(const double &, const float &, const float &)> sum_function;
+    double (*sum_function_ptr)(const double &, const float &, const float &);
 
     if(decay == "exp")
-        sum_function = [](const double &distance, const float &radius, const float &var)
-                       { return exp(-1*distance/radius) * var; };
+        sum_function_ptr = &exp_decay;
     if(decay == "linear")
-    	sum_function = [](const double &distance, const float &radius, const float &var)
-                       { return (1.0-distance/radius) * var; };
+        sum_function_ptr = &linear_decay;
     if(decay == "flat")
-        sum_function = [](const double &distance, const float &radius, const float &var)
-                       { return var; };
+        sum_function_ptr = &flat_decay;
 
     for (int i = 0 ; i < distances.size() ; i++) {
         int nodeid = distances[i].first;
@@ -371,7 +384,7 @@ Accessibility::aggregateAccessibilityVariable(
 
         for (int j = 0 ; j < vars[nodeid].size() ; j++) {
             cnt++;  // count items
-            sum += sum_function(distance, radius, vars[nodeid][j]);
+            sum += (*sum_function_ptr)(distance, radius, vars[nodeid][j]);
 
             // stddev is always flat
             sumsq += vars[nodeid][j] * vars[nodeid][j];