cameron.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
govg has quit [Remote host closed the connection]
govg has joined #mlpack
< jenkins-mlpack>
Starting build #2157 for job mlpack - svn checkin test (previous build: SUCCESS)
nullptr has quit [Read error: Connection reset by peer]
< jenkins-mlpack>
Starting build #2158 for job mlpack - svn checkin test (previous build: SUCCESS)
govg has joined #mlpack
islamfaisal has quit [Ping timeout: 272 seconds]
islamfaisal has joined #mlpack
< zoq>
The to_string_test.cpp does not compile any more with clang -std=c++11 on OS X 10.10, I think because implicit conversion to void* of ostringstream has been removed.
< zoq>
As pointed out in the error message it's 'testOstream << d;' that produced the error. I'm not sure what this line tests anyway, maybe some can clarify?
< jenkins-mlpack>
* Ryan Curtin: If gradient2 or gradient1 are zero, then BOOST_REQUIRE_CLOSE will fail, so use
< jenkins-mlpack>
BOOST_REQUIRE_SMALL in those situations.
< jenkins-mlpack>
* Ryan Curtin: strlen() returns the length of the string but you must account for the null
< jenkins-mlpack>
terminator yourself. Hence, this code sometimes caused random invalid writes
< jenkins-mlpack>
and crashes.
< jenkins-mlpack>
* Ryan Curtin: Remove debugging random seed.
< jenkins-mlpack>
Starting build #2160 for job mlpack - svn checkin test (previous build: FAILURE -- last SUCCESS #2158 2 hr 27 min ago)
< naywhayare>
zoq: I'll take a look into that, hang on
< naywhayare>
zoq: that particular line tests whether or not mlpack objects with the ToString() method can be sent to ostringstream objects
< naywhayare>
theoretically the file src/mlpack/core/util/ostream_extra.hpp should provide this support
< naywhayare>
but I'm not sure why that isn't working with clang. let me test on Linux with clang
< zoq>
I think it worked with an older version of clang.
< naywhayare>
works on debian with clang 3.5.0-6
< naywhayare>
what version are you using?
< zoq>
"Apple LLVM version 6.0 (clang-600.0.54) (based on LLVM 3.5svn)"
< naywhayare>
ok; let me see if I can find a newer version...
< naywhayare>
nope, nothing newer in the debian sid repos
< zoq>
hm, okay
< naywhayare>
if you comment out that line, does compilation still fail? i.e. is it DiscreteDistribution<> that is causing the failure, or is the operator<<() in ostream_extra.hpp not being overloaded for anything?
< zoq>
I need to comment out every "testOstream << d;" so it's not DiscreteDistribution that is causing the failure. Beside that there is another failure in the gmm_main.cpp file that breaks the build ... but if I comment out that line (gmm) everything is fine
< naywhayare>
this is why I need to get an OS X VM running and stable :(
< zoq>
Everything was fine before OS X 10.10 :(
< naywhayare>
it's possible that ostream_extra.hpp contains code that isn't actually valid C++11
< naywhayare>
but I'll have to research it further
< naywhayare>
hm, what happens when you compile with -std=gnu++11 instead of -std=c++11? you can just modify line 42 in CMake/CXX11.cmake
< naywhayare>
actually I wonder if the use of -std=c++0x is causing a problem there, but that should be equivalent to -std=c++11 (at least, it is on gcc)
< naywhayare>
I can't make any sense of that GMM error... the call matches the KMeans constructor perfectly; I don't know why it wants to cast anything to 'const double'
< zoq>
yeah, it makes no sense, std=gnu++11 did not help
< naywhayare>
hm, ok; I'll start trying to get my VM (which I haven't fired up in ages...) updated to OS X 10.10
< naywhayare>
maybe I can set it up as a Jenkins slave too...