verne.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/
Nilabhra has joined #mlpack
tham has joined #mlpack
< uzipaz> tham: I checked your blog post, unfortunately I couldn't find that bit where we specify path, anyways, I can specify the path manually when using cmake and it works, but now, I am getting linking errors when compiling, you can see the error here, http://pastebin.com/FAFsKrPU
< rcurtin> uzipaz: if you have time, could you try with mlpack-2.0.1 instead of the mlpack git master branch?
< uzipaz> tham: I've been told that its probably because of multiple boost installation and I should install boost with static libs, i will try that soon and see how that works
< rcurtin> we made some changes to how the windows build works recently; I am wondering if that could be part of your issue
< uzipaz> rcurtin: certainly, ill let you know how that works out :)
tham has quit [Ping timeout: 250 seconds]
tham has joined #mlpack
< tham> uzipaz : You can find out how to setup the path of boost in step 10 of my blog
< tham> I put the file on pastebin(http://pastebin.com/hgBXvvhq)
< tham> start from line 66~87 as my blog said
< tham> From the error log, it looks like you do not link to the boost libraries
< tham> If you specify the boost path correct
< tham> CMake should link to boost lib automatic
< tham> In the worst case, you need to specify which boost lib you need to link
< tham> you can start from absolute path first, after it work, change them to dependent path if you like
< tham> you can link to specific boost lib like this
< tham> target_link_libraries(${PROJECT_NAME} c:/lib/boost/boost_1_60_0/x64/libboost_date_time${BOOST_LIB_SUFFIX})
< uzipaz> tham: Thanks, ill try it and let you know
Nilabhra has quit [Remote host closed the connection]
< tham> uzipaz : you are welcome, hope you success
tham has quit [Quit: Page closed]
Nilabhra has joined #mlpack
Mathnerd314 has quit [Ping timeout: 268 seconds]
Nilabhra has quit [Remote host closed the connection]
uzipaz has quit [Ping timeout: 250 seconds]
Nilabhra has joined #mlpack
< zoq> tham: I don't think there is any need to specify the boost/armadillo path inside the CMake file. You can just use BOOST_INCLUDEDIR and BOOST_LIBRARYDIR, as uzipaz did.
mentekid_ has joined #mlpack
mentekid has quit [Ping timeout: 244 seconds]
Nilabhra has quit [Remote host closed the connection]
tham has joined #mlpack
< tham> zoq : Do you mean the installer?
< tham> But without setting the path, like BOOST_INCLUDEDIR, BOOST_ROOT and others
< tham> cmake cannot find those libs, unless they are place at the path CMakeLists.txt specify
< tham> You prefer user input the value from command line?
< tham> I suggest change absolute path to dependent path because
< tham> users may not want to install their libs in c drive(like me)
< tham> However, make the path become dependent or not is not a big deal
< tham> The problem need to solve are the links of the dependencies
< tham> Will they always the same?If not, we need someway to tell the installer, where could it find the new link
Mathnerd314 has joined #mlpack
Nilabhra has joined #mlpack
Nilabhra has quit [Ping timeout: 260 seconds]
u8sand_ has joined #mlpack
< zoq> tham: Actually I meant uzipaz's problem, which isn't that cmake can't find boost.
u8sand has quit [Ping timeout: 264 seconds]
Nilabhra has joined #mlpack
< tham> zoq : you are right, my bad, It is not about the problem of cannot find boost
< tham> uzipaz : sorry, I misunderstand your problem, it is not about you did not link to boost, but
< tham> maybe your libs are not compile with same options, different arch, stack overflow has a post discuss about this issue(http://stackoverflow.com/questions/14181351/i-got-duplicate-section-errors-when-compiling-boost-regex-with-size-optimizati#)
< tham> uzipaz is offline, maybe he would check the log
< rcurtin> yeah, I think uzipaz checks the logs, he's said so before
< tham> zoq : Study selective search and edgeBoxes
< tham> I think selective search is easier to implement than edgeBoxes
< tham> But edgeBoxes run faster, and has similar performance compare with selective search
tham has quit [Quit: Page closed]
< zoq> tham: I agree, selective search is easier to implement. However, since we could use the Hoeffding tree or parts of the implementation it shouldn't be too hard to implement edgeBoxes as well. Anyway, let's see what Nilay likes to implement.
< zoq> tham: I hope, if uzipaz rebuilds boost with '--build-type=complete' we can get rid of all the errors.
gtank has quit [Ping timeout: 244 seconds]
gtank has joined #mlpack
Nilabhra has quit [Remote host closed the connection]