R package compiles

[tdhock]

@aldro61 these changes were required to make the new code compile on my computer... however now I am getting a segfault when I run the R package tests...

[tdhock]

on master I got the following compilation errors

thocking@silene:~/projects/mmit(master*)$ python setup.py install --prefix=$HOMErunning install
running bdist_egg
running egg_info
writing mmit.egg-info/PKG-INFO
writing top-level names to mmit.egg-info/top_level.txt
writing dependency_links to mmit.egg-info/dependency_links.txt
reading manifest file 'mmit.egg-info/SOURCES.txt'
writing manifest file 'mmit.egg-info/SOURCES.txt'
installing library code to build/bdist.linux-x86_64/egg
running install_lib
running build_py
creating build
creating build/lib.linux-x86_64-2.7
creating build/lib.linux-x86_64-2.7/mmit
copying mmit/__init__.py -> build/lib.linux-x86_64-2.7/mmit
creating build/lib.linux-x86_64-2.7/mmit/core
copying mmit/core/__init__.py -> build/lib.linux-x86_64-2.7/mmit/core
creating build/lib.linux-x86_64-2.7/mmit/tests
copying mmit/tests/trivial_tests.py -> build/lib.linux-x86_64-2.7/mmit/tests
copying mmit/tests/__init__.py -> build/lib.linux-x86_64-2.7/mmit/tests
running build_ext
building 'mmit.core.solver' extension
creating build/temp.linux-x86_64-2.7
creating build/temp.linux-x86_64-2.7/mmit
creating build/temp.linux-x86_64-2.7/mmit/core
gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -fPIC -I/usr/include/python2.7 -I/usr/local/lib/python2.7/dist-packages/numpy/core/include -c mmit/core/solver_python_bindings.cpp -o build/temp.linux-x86_64-2.7/mmit/core/solver_python_bindings.o -std=c++0x
cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for Ada/C/ObjC but not for C++ [enabled by default]
In file included from mmit/core/piecewise_function.h:22:0,
                 from mmit/core/solver.h:19,
                 from mmit/core/solver_python_bindings.cpp:6:
mmit/core/coefficients.h:21:24: sorry, unimplemented: non-static data member initializers
mmit/core/coefficients.h:21:24: error: ‘constexpr’ needed for in-class initialization of static data member ‘quadratic’ of non-integral type
mmit/core/coefficients.h:22:21: sorry, unimplemented: non-static data member initializers
mmit/core/coefficients.h:22:21: error: ‘constexpr’ needed for in-class initialization of static data member ‘linear’ of non-integral type
mmit/core/coefficients.h:23:23: sorry, unimplemented: non-static data member initializers
mmit/core/coefficients.h:23:23: error: ‘constexpr’ needed for in-class initialization of static data member ‘constant’ of non-integral type
mmit/core/coefficients.h: In constructor ‘Coefficients::Coefficients()’:
mmit/core/coefficients.h:25:26: error: ‘class Coefficients’ has no member named ‘quadratic’
mmit/core/coefficients.h:25:47: error: ‘class Coefficients’ has no member named ‘linear’
mmit/core/coefficients.h:25:65: error: ‘class Coefficients’ has no member named ‘constant’
mmit/core/coefficients.h: In constructor ‘Coefficients::Coefficients(double, double, double)’:
mmit/core/coefficients.h:26:54: error: ‘class Coefficients’ has no member named ‘quadratic’
mmit/core/coefficients.h:26:75: error: ‘class Coefficients’ has no member named ‘linear’
mmit/core/coefficients.h:26:93: error: ‘class Coefficients’ has no member named ‘constant’
In file included from mmit/core/solver.h:19:0,
                 from mmit/core/solver_python_bindings.cpp:6:
mmit/core/piecewise_function.h: At global scope:
mmit/core/piecewise_function.h:37:36: error: invalid use of non-static data member ‘PiecewiseFunction::breakpoint_coefficients’
mmit/core/piecewise_function.h:41:56: error: from this location
mmit/core/piecewise_function.h:41:84: sorry, unimplemented: non-static data member initializers
mmit/core/piecewise_function.h:41:84: error: in-class initialization of static data member ‘min_ptr’ of non-literal type
error: command 'gcc' failed with exit status 1
thocking@silene:~/projects/mmit(master*)$ 

[tdhock]

I got the following when I ran the R package tests using the code on this branch

bash R_CMD_build_check.sh 
Warning: invalid uid value replaced by that for user 'nobody'

* using log directory ‘/home/thocking/projects/mmit/MMIT.Rcheck’
* using R version 3.3.3 (2017-03-06)
* using platform: x86_64-pc-linux-gnu (64-bit)
* using session charset: UTF-8
* checking for file ‘MMIT/DESCRIPTION’ ... OK
* this is package ‘MMIT’ version ‘2017.03.15’
* checking package namespace information ... OK
* checking package dependencies ... OK
* checking if this is a source package ... OK
* checking if there is a namespace ... OK
* checking for executable files ... OK
* checking for hidden files and directories ... OK
* checking for portable file names ... OK
* checking for sufficient/correct file permissions ... OK
* checking whether package ‘MMIT’ can be installed ... OK
* checking installed package size ... OK
* checking package directory ... OK
* checking DESCRIPTION meta-information ... OK
* checking top-level files ... OK
* checking for left-over files ... OK
* checking index information ... OK
* checking package subdirectories ... OK
* checking R files for non-ASCII characters ... OK
* checking R files for syntax errors ... OK
* checking whether the package can be loaded ... OK
* checking whether the package can be loaded with stated dependencies ... OK
* checking whether the package can be unloaded cleanly ... OK
* checking whether the namespace can be loaded with stated dependencies ... OK
* checking whether the namespace can be unloaded cleanly ... OK
* checking dependencies in R code ... OK
* checking S3 generic/method consistency ... OK
* checking replacement functions ... OK
* checking foreign function calls ... OK
* checking R code for possible problems ... OK
* checking Rd files ... OK
* checking Rd metadata ... OK
* checking Rd cross-references ... OK
* checking for missing documentation entries ... OK
* checking for code/documentation mismatches ... OK
* checking Rd \usage sections ... OK
* checking Rd contents ... OK
* checking for unstated dependencies in examples ... OK
* checking line endings in C/C++/Fortran sources/headers ... OK
* checking line endings in Makefiles ... OK
* checking compilation flags in Makevars ... OK
* checking for GNU extensions in Makefiles ... OK
* checking for portable use of $(BLAS_LIBS) and $(LAPACK_LIBS) ... OK
* checking compiled code ... NOTE
File ‘MMIT/libs/MMIT.so’:
  Found ‘_ZSt4cout’, possibly from ‘std::cout’ (C++)
    Object: ‘piecewise_function.o’

Compiled code should not call entry points which might terminate R nor
write to stdout/stderr instead of to the console, nor the system RNG.

See ‘Writing portable packages’ in the ‘Writing R Extensions’ manual.
* checking examples ... ERROR
Running examples in ‘MMIT-Ex.R’ failed
The error most likely occurred in:

> ### Name: compute_optimal_costs
> ### Title: compute optimal costs
> ### Aliases: compute_optimal_costs
> 
> ### ** Examples
> 
> library(MMIT)
> target.mat <- rbind(
+   c(-1, Inf),
+   c(-2, 3),
+   c(-Inf, 1))
> input.vec <- c(-2.2, 3, 10)
> compute_optimal_costs(input.vec, target.mat, 0)

INSERTION COMPLETED
----------------------------
N pointer moves: 0

 *** caught segfault ***
address (nil), cause 'memory not mapped'

Traceback:
 1: .C("compute_optimal_costs_interface", n_data = length(feature.vec),     feature_vec = as.double(feature.vec), lower_vec = as.double(lower.vec),     upper_vec = as.double(upper.vec), margin = as.double(margin),     loss = as.integer(fun.vec[[loss]]), moves_vec = as.integer(neg.vec),     pred_vec = as.double(inf.vec), cost_vec = as.double(inf.vec),     NAOK = TRUE, PACKAGE = "MMIT")
 2: compute_optimal_costs(input.vec, target.mat, 0)
An irrecoverable exception occurred. R is aborting now ...
Segmentation fault (core dumped)
* checking for unstated dependencies in ‘tests’ ... OK
* checking tests ...
  Running ‘testthat.R’
 ERROR
Running the tests in ‘tests/testthat.R’ failed.
Last 13 lines of output:
   4: eval(exprs, env)
   5: source_file(path, new.env(parent = env), chdir = TRUE)
   6: force(code)
   7: with_reporter(reporter = reporter, start_end_reporter = start_end_reporter,     {        lister$start_file(basename(path))        source_file(path, new.env(parent = env), chdir = TRUE)        end_context()    })
   8: FUN(X[[i]], ...)
   9: lapply(paths, test_file, env = env, reporter = current_reporter,     start_end_reporter = FALSE, load_helpers = FALSE)
  10: force(code)
  11: with_reporter(reporter = current_reporter, results <- lapply(paths,     test_file, env = env, reporter = current_reporter, start_end_reporter = FALSE,     load_helpers = FALSE))
  12: test_files(paths, reporter = reporter, env = env, ...)
  13: test_dir(test_path, reporter = reporter, env = env, filter = filter,     ...)
  14: with_top_env(env, {    test_dir(test_path, reporter = reporter, env = env, filter = filter,         ...)})
  15: run_tests(package, test_path, filter, reporter, ...)
  16: test_check("MMIT")
  An irrecoverable exception occurred. R is aborting now ...
  Segmentation fault (core dumped)
* checking PDF version of manual ... OK
* DONE

Status: 2 ERRORs, 1 NOTE
See
  ‘/home/thocking/projects/mmit/MMIT.Rcheck/00check.log’
for details.

make: *** [Rpackage.check] Error 1

[aldro61]

I deleted my previously compiled version and trie to recompile with the new code (this branch) and I'm also getting a segfault. Don't know whats wrong. I'll investigate.

[aldro61]

I found a clue. The first time insert_point is called, the following condition is false, but it should be true (and is true with the code on master):

this->min_ptr == this->breakpoint_coefficients.end()

Also, if I verify this condition inside the constructor of PiecewiseFunction, it is true. It's as if the value was changing before we first call insert_point.

[tdhock]

I added some code for automatic testing using https://travis-ci.org/ -- can you please log in to travis and turn it on for this repository?

[tdhock]

not sure what the problem is, but you may want to use a typedef to clarify what is going on here

    // Insert the new breakpoint
    std::pair<std::map<double, Coefficients>::iterator, bool> insert = this->breakpoint_coefficients.insert(std::pair<double, Coefficients>(breakpoint_position, new_breakpoint_coefficients));

[aldro61]

I fixed the segfault (see this commit). It turns out that calling the constructor from within the more simple constructor was messing up the initialization of min_ptr.

[tdhock]

is it OK with you to stop with an error when the features are not sorted? (otherwise I guess the features are not used at all and we could just not pass them at all to the C++ code?)

Before merging this branch with master, maybe we should add some tests with tied features?

[tdhock]

actually, whether or not there are ties in the features won't affect the solver, right? I'm thinking that we should probably just not pass the features at all to the C++ code. What do you think?

[aldro61]

I think that features could simply not be passed to the solver. What the code is really doing is adding one point at a time. The outputted values also correspond to each point.

Update: yup, I agree.

I found a new test that fails. I’ll fix it and add it to the tests.

[tdhock]

ok great I will remove that new test then

[tdhock]

ok I removed the feature_vec argument from the C++ solver function

[aldro61]

We should add a test that checks that the total cost/prediction (last value of the returned arrays) are the same for different shufflings of points.