hkaiser changed the topic of #ste||ar to: STE||AR: Systems Technology, Emergent Parallelism, and Algorithm Research | stellar.cct.lsu.edu | HPX: A cure for performance impaired parallel applications | github.com/STEllAR-GROUP/hpx | Buildbot: http://rostam.cct.lsu.edu/ | Log: http://irclog.cct.lsu.edu/
jaafar has quit [Ping timeout: 255 seconds]
K-ballo has quit [Quit: K-ballo]
eschnett_ has joined #ste||ar
khuck has joined #ste||ar
hkaiser has quit [Quit: bye]
khuck has quit [Read error: Connection reset by peer]
khuck_ has joined #ste||ar
eschnett_ has quit [Quit: eschnett_]
Abhishek09 has joined #ste||ar
khuck_ has quit [Remote host closed the connection]
nikunj97 has joined #ste||ar
jaafar has joined #ste||ar
khuck has joined #ste||ar
Abhishek09 has quit [Ping timeout: 256 seconds]
Abhishek09 has joined #ste||ar
<Abhishek09>
hello guys
<Abhishek09>
Anyone who used docker?
jaafar has quit [Ping timeout: 258 seconds]
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
Abhishek09 has quit [Ping timeout: 256 seconds]
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Ping timeout: 255 seconds]
nikunj has joined #ste||ar
david_pfander has joined #ste||ar
daissgr has joined #ste||ar
daissgr has quit [Ping timeout: 250 seconds]
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
Abhishek09 has joined #ste||ar
nikunj has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
daissgr has joined #ste||ar
<Abhishek09>
hi diehlpk_work are you here?
<simbergm>
Abhishek09: he's usually sleeping at this time... (he's in some us time zone)
<Abhishek09>
hey simbergm is there any idea about manyliinux?
<simbergm>
no, sorry, never heard of that
<Abhishek09>
you know who participating this year in gsoc
nikunj has quit [Ping timeout: 272 seconds]
nikunj has joined #ste||ar
nikunj1997 has joined #ste||ar
nikunj97 has quit [Ping timeout: 268 seconds]
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
<zao>
manylinux1 is fun.
<zao>
I looked into this briefly when looking at deploying my library for operating on Path of Exile game data files as a binary Python library.
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
<zao>
Abhishek09: A common way of building manylinux1 wheel as I assume you've found is indeed to boot up a Docker container that has ancient enough compilers installed and build it inside there.
<zao>
I'm suspecting that the C++ compiler and libstdc++ for manylinux1 is waaay too old for HPX.
<zao>
I know I ruled it out for my rather vanilla C++.
<zao>
But it doesn't hurt to figure out what it can actually do.
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
heller_ has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
K-ballo has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Client Quit]
nikunj has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Read error: Connection reset by peer]
nikunj has joined #ste||ar
nikunj has quit [Client Quit]
K-ballo has quit [Ping timeout: 246 seconds]
nikunj has joined #ste||ar
nikunj1997 has quit [Ping timeout: 255 seconds]
adityaRakhecha has joined #ste||ar
daissgr has joined #ste||ar
hkaiser has joined #ste||ar
adityaRakhecha has quit [Ping timeout: 256 seconds]
adityaRakhecha has joined #ste||ar
<adityaRakhecha>
Hello to all. I am building project phylanx, are Interpreter tests neccesary for pybind11 ?
jaafar has joined #ste||ar
khuck has quit [Read error: Connection reset by peer]
khuck has joined #ste||ar
eschnett_ has joined #ste||ar
eschnett_ has quit [Quit: eschnett_]
K-ballo has joined #ste||ar
<hkaiser>
pybind11 is necessary for the interpreter tests, yes
K-ballo has quit [Read error: Connection reset by peer]
K-ballo has joined #ste||ar
<Yorlik>
hkaiser: I think new_ doesn't work with hpx::launch::sync, I have to use local_new. At least I didn't get it to work and just used local_new. What I found peculiar though was, that local_new still returns a future and requires me to use future.get() even in the sync case - I think this should be handled like with the shared pointers where in the sync case you directly obtain the pointer. This indirection of
<Yorlik>
local_new shouldn't be necessary, imho.
<K-ballo>
we have local_new? interesting
aserio has joined #ste||ar
<Yorlik>
Yes, but it returns a future :D
<Yorlik>
A very immediate one though ...
<K-ballo>
are you saying it is synchronous?
<K-ballo>
returns an already ready future?
<Yorlik>
heller- told me so yesterday
<zao>
Easier to compose than if you had to wrap it yourself, maybe?
<Yorlik>
I don't know if it's just sugar for future.get() or a real sync call
<Yorlik>
After all you can't avoid to have the instance register with agas and all that boilrplate stuff.
<Yorlik>
maybe the roundtrip is slightly faster - really I donÄt know.
<zao>
Used make_ready_future back when new, at least.
aserio has quit [Ping timeout: 264 seconds]
<K-ballo>
supports non-copyable and non-movable arguments, must be immediate
<K-ballo>
at least the object creation part must be
<Yorlik>
How would you sync create an instance?
eschnett_ has joined #ste||ar
<K-ballo>
an instance of what?
<Yorlik>
Of a component
<K-ballo>
why? (matters to what the answer is)
<hkaiser>
Yorlik: you're right, I apologize
<Yorlik>
There might be object you mainly use locally, but which will receive occasional remote calls. You create them locally, mostly use them locally.
<K-ballo>
and how is that related to sync?
<hkaiser>
new_() does not accept a launch policy (not sure why I was thinking it would - I might have just planned to add it)
<zao>
There's manylinux1 wheels for whatever pypi considers "cryptography"
<zao>
At least for some Python versions.
<diehlpk_work>
Abhishek09, Yes, manylinux uses Cent OS 5
<diehlpk_work>
The RHEL (and CentOS) 5.x series use GCC 4.1 and do not provide newer versions because they are stable, "enterprise" distributions so you can't update to a newer version using yum
<diehlpk_work>
HPX would not compile with GCC 4.1 and needs GCC 4.9 or newer
K-ballo has joined #ste||ar
<Abhishek09>
gcc 4.8 possible?
<diehlpk_work>
Abhishek09, Therefore my question again, have you ever compiled hpx and phylanx
<diehlpk_work>
let me do the math for you 4.8 < 4.9 and 4.9 is the minimal working version
<zao>
I'd recommend building things manually on a regular modern Linux machine to get a feeling for how it should work, and then try doing the same in the pypa docker container.
<diehlpk_work>
So the math say you can not use 4.8
<parsa>
zao: i don't quite get the incentive to do static builds here. if someone has an old Linux it won't work anyway
<Abhishek09>
no, because it require linux and window
<zao>
parsa: How so? If you build the new compiler and its stdlib inside an old environment, it'll depend on old glibc.
<diehlpk_work>
Abhishek09, As I told you before, it would be sufficient to get a pip package working on one OS
<Abhishek09>
only slight lower than 4.9
<zao>
And if you privately deploy the runtimes or static link, there won't be much in the way of conflicts.
<parsa>
zao: what! there is no ABI incompatibility between current GCC and GCC 4.1?
<diehlpk_work>
If you can provide us with a pip package for Ubuntu or Fedora, we would be really happy
<zao>
parsa: Well, you're not interoperating any standard C++ library types.
<Abhishek09>
i have mojave thats why i not complile
<parsa>
makes sense now
<daissgr>
parsa: Sort of - the constructor in struct_of_array.hpp converts the AoS array with the expansions and space_vectors into a simple SoA array
<zao>
(private deployment of shared libs may be out of the question, I guess, as the symbols might be picked up)
<parsa>
Abhishek09: then build a pip package for mojave. and use docker if you want to build for other distros once you're done with mojave
<diehlpk_work>
Abhishek09, Yes, I agree with parsa that you get the pip package running on one OS.
<diehlpk_work>
if you are able to get the setup.py and binary wheels done for one machine, it should be easy to migrate them to other modern linux os
<zao>
You'd be producing something akin to "cryptography-2.5-cp34-abi3-macosx_10_6_intel.whl" I guess.
<Abhishek09>
build instructions is not given for macos?why
<parsa>
Abhishek09: you can add the instructions for building on a mac if you want
<parsa>
the package manager would become Homebrew or Port or whatever, and the rest of the commands are probably the same for macOS
jaafar_ has joined #ste||ar
<Abhishek09>
is build instruction same for macos?
<parsa>
i don't think so
<daissgr>
parsa: That's to load on of the AoS vales into the SoA array - it is used to set the boundary values. The constructor at L59 loads more. You don't really need these methods though I think. The original data-structure you are using is in a SoA format already if I recall correctly. You just need to skip the generation of the space_vectors and use the components that were used for the creation instead
<Abhishek09>
have you build on macos before?
<parsa>
haven't built on mac since 6-7 months ago
<parsa>
Abhishek09: why? do you have an issue building on mac?
khuck has quit [Remote host closed the connection]
khuck has joined #ste||ar
khuck has quit [Read error: No route to host]
<adityaRakhecha_>
I am stuck at this one issue. When I give path for a package's folder, there is one file missing which is in another folder of same package and when I give that path, one from the former gets missing. I have tried everything I could but nothing works.
<adityaRakhecha_>
It could be a trivial issue. I accept. But I am doing and learning as you suggested diehlpk_work.
khuck has joined #ste||ar
<diehlpk_work>
adityaRakhecha_, What are you try to do?
<adityaRakhecha_>
In cmake, path for blaze, -Dblaze_DIR=/blaze/install/path/share/blaze/cmake/ I am getting this error `include could not find load file: blaze-config-version.cmake`
<adityaRakhecha_>
And when I give path where this file is located I get error for another file.
khuck has quit [Read error: Connection reset by peer]
khuck has joined #ste||ar
khuck_ has joined #ste||ar
khuck has quit [Ping timeout: 250 seconds]
<hkaiser>
diehlpk_work: have you installed blaze on your machine?
<hkaiser>
adityaRakhecha_: ^^
<diehlpk_work>
hkaiser, Via the fedora package?
<adityaRakhecha_>
hkaiser: yes.
parsa is now known as parsa_
<diehlpk_work>
adityaRakhecha_, You are using the fedora package
parsa_ is now known as parsa
<adityaRakhecha_>
diehlpk_work : No. I am on Ubuntu 16.04
<hkaiser>
diehlpk_work: I meant to talk to adityaRakhecha_, sorry
<hkaiser>
adityaRakhecha_: where did you install blaze to (what directory)?
<adityaRakhecha_>
I have installed all 4 (hpx, blaze, phylanx and pybind11) in same directory.
<diehlpk_work>
adityaRakhecha_, Have you compiled them?
david_pfander has quit [Ping timeout: 246 seconds]
<diehlpk_work>
And installed them?
<diehlpk_work>
hkaiser, Nevertheless, I checked the fedora blaze package and cmake could find it
<hkaiser>
what's the name of the directory you installed things to?
<adityaRakhecha_>
Yes. Only phylanx is giving these error.
<hkaiser>
diehlpk_work: sure
<diehlpk_work>
adityaRakhecha_, Have you run cmake, make, make install
<adityaRakhecha_>
`stellar`
<hkaiser>
adityaRakhecha_: what's the full path of the directory you installed things to?
<hkaiser>
you installed it to /stellar?
<adityaRakhecha_>
Yes. For rest 3 everything worked fine.
<adityaRakhecha_>
yes
<hkaiser>
why do you use this command line option, then: -Dblaze_DIR=/blaze/install/path/share/blaze/cmake/
<adityaRakhecha_>
`/home/aditya/stellar/`
<adityaRakhecha_>
hkaiser: In cmake for phylanx
<hkaiser>
I didn't ask where, I asked why?
<hkaiser>
this option should point to the directory on YOUR disk where blaze was installed
<hkaiser>
i.e. something like /home/aditya/stellar/share/blaze/cmake/
<hkaiser>
please just look around where the blaze-config.cmake file is located, point your command line option to that directory
<adityaRakhecha_>
Yes. I tried this by changing it to the directory of missing file but then it gives error for another file that is missing from other folder of same directory.
<zao>
What file?
<adityaRakhecha_>
`blaze-targets.cmake`
<adityaRakhecha_>
which is in `/home/aditya/stellar/blaze/cmake/`
<zao>
Hmm, is HPX a real and working backend for Blaze threading, or should one aim for OpenMP/C++11?
<zao>
(personal musing)
<hkaiser>
zao: HPX should be working
<hkaiser>
adityaRakhecha_: what other file is it complaining about?
<Abhishek09>
adityaRakhecha_:you working on which project?
<adityaRakhecha_>
In my system `blaze-targets.cmake` is in `/home/aditya/stellar/blaze/build/CMakeFiles/Export/share/blaze/cmake` and `blaze-config-version.cmake` is in `/home/aditya/stellar/blaze/cmake/`
<zao>
What "external" are these, the tree that only has asio's basic_resolver_iterator.hpp and hpx-no-inspect?
<hkaiser>
yes
<daissgr>
parsa: What have you changed in the buildscripts repo? On current master they try to clone and build octotiger in /octotiger. As much as I like octotiger I don't want it in my root directory :p
<zao>
daissgr: Sounds like it's container-aimed :)
<parsa>
daissgr: where is that coming from? i have tested powertiger on krypton and rostam and didn't have that issue
<parsa>
no i've never tested it in a container
<daissgr>
I just updated the powertiger scripts on one of my machines. Maybe the issue is that is was prebuild with an older commit of the buildscripts?
<parsa>
probably not. i haven't changed the directories
<parsa>
same directory structure should work
<Abhishek09>
but someone has told on this you can build on mac
<daissgr>
oh I see what happened! The source-me is not called at the beginning of build-octotiger.sh anymore
<parsa>
do module use ~/powertiger/build/modules(?) and then module load hpx vc
<parsa>
sorry module load hpx silo
<parsa>
daissgr: i removed the source-me from the package builds since it was being called several times in different spots
<daissgr>
ah the INSTALL_ROOT and SOURCE_ROOT variables are now defined there?
<daissgr>
because the behaviour is due to them being empty when I just call build-octotiger
<Abhishek09>
hey diehlpk_work: how to build phylanx on mojave? i facing issue
<daissgr>
I guess I still have to rebuild everything? There is nothing about modules in my build directory
<zao>
Blargh, phylanx still installs an egg into ~/.local/lib
<zao>
I had forgotten it did that :(
aserio has quit [Ping timeout: 250 seconds]
aserio has joined #ste||ar
<Abhishek09>
parsa: would you how to build phylanx on mojave?
<daissgr>
still, why is the equivalent to the source-me.sh (I guess config.sh) not being called at the beginning of the build-octotiger script? I called originally from several places for convienence to avoid loading anything manually
<parsa>
daissgr: understood, didn't know the script had to support building form scratch more than once
<parsa>
daissgr: it's not being called again because it looked redundant to me without context
<parsa>
so i removed it
parsa is now known as parsa_
parsa_ is now known as parsa
<daissgr>
I often change things around in the toolchain, so having the source-me in each of the build scripts is rather useful to me. That's also why I like to have a per-directory install of all the dependencies.
<daissgr>
it's also useful for our university machines since they share the home folder, but I need to differently compiled toolchain for the krypton machine or for the KNL machine
<parsa>
shrug. source config.sh and then ./build-octotiger.sh?
<diehlpk_work>
Abhishek09, What issues you are facing?
<parsa>
daissgr: give me a few minutes. i'll fix that for you
<daissgr>
okay! It's not really pressing for now I can simply use f28374b
<parsa>
daissgr: don't do that
<parsa>
use dcec08eb40524f8a7d8f411f8cf36f9df2e7752f
<Abhishek09>
i not able to install python3 dev on mojave
<parsa>
hpx master doesn't work on POWER9
<parsa>
Abhishek09: do you not use homebrew, port, fink or anything like that on your mac?
<Abhishek09>
i have tried using homebrew saying no formulae found
<zao>
parsa: I'm currently building Phylanx on last generation's MacOS, with homebrew Boost, hwloc, pybind11, Python, and probably more. Probably mismatched w.r.t C++ versions and all.
<zao>
Sloooowly building, but getting somewhere thus far.
<diehlpk_work>
Abhishek09, Sorry, I can not help you with mojave related questions since I do not have a Mac
<daissgr>
parsa: ah it still does not work? Yeah I was using the same hpx commit but adjusted it manually! quite useful to have it
<parsa>
daissgr: hkaiser says it should be fixed, but haven't checked yet so we don't know if it works for us, too
<zao>
Abhishek09: The "python" thing from Homebrew should contain a Python.framework with headers and all.
<zao>
It "should" find them properly if the build just finds the python interpreter, I believe.
<daissgr>
I'll try it out! Not much else I can currently do before the conference call anyway
<daissgr>
by the way, I created a pull request with the most current gpu branch! might be interesting for you
<diehlpk_work>
daissgr, Why do you want to use hpx master?
parsa is now known as parsa_
<diehlpk_work>
Is the current working commit not sufficient?
<diehlpk_work>
daissgr, Do you know why the file system on krypton is so slow?
<Yorlik>
Seems optimizing computations in terms of granularity and optimal CPU utilization and cache friendlyness is coming down to a multidimensional knapsack problem if you want to synchronize in fixed intervals, like a frame in a game.
<daissgr>
diehlpk_work: what folder are you using?
<Yorlik>
Means - I'm getting into NP hard territory in zero time ;)
<diehlpk_work>
daissgr, My home
<daissgr>
The home is shared between all machines out our Institute (and has a copy-on-write backup I think). Try using /scratch
<daissgr>
simply make your own folder there
<daissgr>
at our institute*
<diehlpk_work>
Ok. And how can I change my password?
<diehlpk_work>
It seems it is only working when I am in your network
<diehlpk_work>
Should I keep my pw
<heller_>
hkaiser: btw, I'm on the train right now and won't make it to the call tonight
<daissgr>
uhm it's been since I have done that. Doesn't passwd simply work?
khuck_ has quit []
<diehlpk_work>
The only supported way to change the password for IPVS accounts is Web based.
<daissgr>
as for your question. While octotiger seems to be working on Krypton there are a few oddities. I want to see whether they still persist with the most current hpx
<daissgr>
for example, there is the crash at initialisation when using to many commandline arguments
<daissgr>
though I have not yet bothered to really look at it and instead just used scenario files to reduce the number of arguments
<daissgr>
besides, I dislike being stuck at a certain commit :p
<zao>
Judging by a successful installation on Linux, it's bailing on the Python packaging bits.
<diehlpk_work>
Abhishek09, You could use a virtual machine or docker on Mac OS
<diehlpk_work>
I am not sure if anyone compiled phylanx on Mac
<parsa_>
zao: the installation script bound together with duct type... someone should fix it. yes it always installs to ~/.local
<zao>
diehlpk_work: Considering how my MacOS build broke mysteriously, that'd probably be a good idea.
<parsa_>
zao: it's hardcored
parsa_ is now known as parsa
* zao
notices the "someone" word and dodges it
iti_ has joined #ste||ar
<diehlpk_work>
daissgr, I am fine without changing my pw, as long as your admins are fine too
<diehlpk_work>
Abhishek09, Could you use Fedora 29, so I can assist your more easy
<parsa>
daissgr: okay, so doing that seems to be difficult. what i'll do instead is add a `-c <target>` flag to build-all.sh that allows you to do stuff like ./build-all <other args> -c octotiger. is that good enough?
nimishkhurana[m] has joined #ste||ar
aserio has quit [Ping timeout: 268 seconds]
<daissgr>
parsa: Sure, that should work for me! Maybe add a usage comment about it in the readme
aserio has joined #ste||ar
<parsa>
daissgr: actually i'll just make it the third argument. so build-all <build type> <cuda support> <target>. will add to readme and synopsis
<daissgr>
okay
<hkaiser>
aserio: yt?
<aserio>
hkaiser: yep
<hkaiser>
aserio: the meeting link for OB today you sent yesterday is wrong
<hkaiser>
it points to the Phylanx meeting
<iti_>
@dielpl_work: Hey, is it good time to talk? Last time I came I guess you weren't here :)
<aserio>
huh
<aserio>
I thought I had fixed that
<daissgr>
aserio: Yep I am seeing Phylanx meeting as well! Is the number still correct?
<hkaiser>
what number?
<daissgr>
the meeting number
<hkaiser>
shrug
<daissgr>
since I always join the meeting in webex by number
<nimishkhurana[m]>
Hye everyone I am Nimish Khurana from India.I am interested in participating in GSoC in Stellar organization.I have gone through proposed ideas and would like to contribute to Phylanx project.I am proficient in Python and Machine Learning.Two projects which i would like to know more about and how to start contributing Test Framework for Phylanx Algorithms and pip package for Phylanx.Can anyone please guide me how
<nimishkhurana[m]>
to get started on these.
<hkaiser>
thanks
<zao>
Hi!
* zao
pokes diehlpk_work
<hkaiser>
zao: he's in a meeting right now
<zao>
Ah, OK.
<zao>
nimishkhurana[m]: For most projects we recommend getting started by building HPX and Phylanx. For the testing one, do contact diehlpk_work on the email listed so they know you're interested, I think.
<zao>
(if I remember correctly)
Abhishek09 has quit [Ping timeout: 256 seconds]
<iti_>
@zao: I was able to build hpx(Although there are few things I need to discuss regarding the project I am interested in). I guess I'll email him then. Thanks for letting know.
Abhishek09 has joined #ste||ar
iti_ has quit [Quit: Leaving]
<nimishkhurana[m]>
ok thanks @za
<nimishkhurana[m]>
OK thanks zao I will build HPX and Phylanx .
<zao>
Regarding the projects themselves, ask about and figure out what the goals are, and prepare a solid application on how you intend to approach them I guess.
<zao>
The whole "building" part tends to be a required part whatever the project is :)
<diehlpk_work>
nimishkhurana[m], I am back
<nimishkhurana[m]>
Yes I understood @zao
<diehlpk_work>
For the Phylanx test Framwork, please contact the mentors listed at the wiki
<diehlpk_work>
I have not much knowledge about that one
<nimishkhurana[m]>
And for pip package one?
<diehlpk_work>
I am happy to help
<nimishkhurana[m]>
That looks quite interesting to me
<nimishkhurana[m]>
Ok.Can I please get some starting pointers.I didnt find any existing work or issue on that.
<nimishkhurana[m]>
I googled on how to make a project installable through pip. Can I follow their official documentation.That looks simple to me
<diehlpk_work>
nimishkhurana[m], The difficult part on this project is that most of the dependencies are not available within pip
<diehlpk_work>
A good start to understand which kind of dependencies are needed would be to compile hpx and phylanx on your local machine
<diehlpk_work>
Somehow you would need to provide binary wheels for some deps
<diehlpk_work>
For example pybind is available in pip
<diehlpk_work>
But not hpx, boost, hwloc, and others
<zao>
Can you even use pip to package non-Python software?
<zao>
Some guides I see seem to suggest "repairing" wheels with externally built shared libraries.
<zao>
(I don't know how wheels work, we build all our stuff separately and leverage PYTHONPATH :D)
<zao>
(we as in my site)
<nimishkhurana[m]>
Ok so the main task is to think about dependencies which are not available through pip.I will build hpx and phylanx first soon.
<diehlpk_work>
I assume they would need to ship mpi
<zao>
diehlpk_work: That's still Python bindings to native code, isn't it?
<zao>
I would guess that the wheels ship some particular MPI installation like OpenMPI, maybe?
<zao>
Ah, nevermind. The wheels are all Windows-based.
<zao>
mpi4py explicitly mandates that you have a MPI compiler wrapper and MPI libraries on the system when building it.
<zao>
The understanding I have regarding native libraries is that a binary wheel can contain shared libraries, but they're for binding inside the wheel itself, not directly usable from another package. I could be wrong on that tho.
<zao>
Heh, Windows version of mpi4py is built against "msmpi.dll".
<zao>
That still just talks about extension modules, doesn't it?
<zao>
While you might be able to ship a HPX installation with headers and libraries in a wheel, would anything be able to use that, and would it be acceptable to repositories?
<diehlpk_work>
Maybe not, but having hpx compiled could work
<zao>
The workflow I seem to see out there is to have everything separately built, build Phylanx into a wheel referencing all those libraries, and then `auditwheel repair`.
<zao>
Time to hit the sack, good night all.
<zao>
I guess some of this is up to the GSoC people to find out the fun way :)
<diehlpk_work>
zao, Part of this project is that the student investigates how to build a pip
<diehlpk_work>
Yes, good night
<diehlpk_work>
Maybe the student comes up with it is not possible
<Abhishek09>
hey diehlpk_work: audit wheel is used by manylinux which is not possible.
daissgr has quit [Ping timeout: 264 seconds]
aserio has quit [Ping timeout: 245 seconds]
aserio has joined #ste||ar
daissgr has joined #ste||ar
Abhishek09 has quit [Quit: Page closed]
Abhishek09 has joined #ste||ar
<Abhishek09>
hey getting the error - Could not find a package configuration file provided by "HPX" with any of the following names: HPXConfig.cmake hpx-config.cmake
<hkaiser>
Abhishek09: have you followed the install instructions for HPX?
<hkaiser>
Abhishek09: this is while trying to build Phylanx, right?
<Abhishek09>
yes
aserio has quit [Quit: aserio]
<hkaiser>
Abhishek09: 'yes' to the first or the second question
<Abhishek09>
i am trying to install blaze
<Abhishek09>
i have installed hpx
<hkaiser>
where did you install it to (what path)?
<hkaiser>
(i.e. what CMAKE_INSTALL_PREFIX did you specify while configuring HPX)
<Abhishek09>
home/hpx
<Abhishek09>
macintosh/hpx
<Abhishek09>
sorry
<hkaiser>
ok, somewhere underneath that directory should be a cmake/HPX/ folder that has the HPXCOnfig.cmake and friends
<Abhishek09>
other in Macintosh ▸ hpx ▸ install ▸ path ▸ lib ▸ cmake ▸ HPX
<Abhishek09>
hey hkaiser are you here
<hkaiser>
wel, so you have that directory
<hkaiser>
now point blaze to that directory
<hkaiser>
-DHPX_DIR=....
<Abhishek09>
which one
<hkaiser>
the one you installed HPX to ;-)
nikunj has quit [Ping timeout: 245 seconds]
<Abhishek09>
i have 2 folder named hpx
<hkaiser>
but only one folder where you installed HPX to, the other one is where you built it
<Abhishek09>
inside hpx another hpx exist
<hkaiser>
that's where you built it, no?
<Abhishek09>
i have sucessfuully installed hpx
<hkaiser>
Abhishek09: so the bottom line is that your initial comment was your cmake was not able to find the HPXConfig.cmake file, and now you're telling me that you found it - good
<hkaiser>
now the only thing left is to tell Blaze where you found that file
<Abhishek09>
i have 3 location see that
<hkaiser>
Abhishek09: use the one you specified as the install prefix while configuring HPX
daissgr has quit [Quit: WeeChat 1.9.1]
<Abhishek09>
By not providing "FindHPX.cmake" in CMAKE_MODULE_PATH this project has asked CMake to find a package configuration file provided by "HPX", but CMake did not find one. Could not find a package configuration file provided by "HPX" with any of the following names: HPXConfig.cmake hpx-config.cmake
<Abhishek09>
Add the installation prefix of "HPX" to CMAKE_PREFIX_PATH or set "HPX_DIR" to a directory containing one of the above files. If "HPX" provides a separate development package or SDK, be sure it has been installed. -- Configuring incomplete, errors occurred!
<Abhishek09>
hkaiser again we have same error!
<hkaiser>
Abhishek09: have you specified the full path of the directory where the HPXConfig.cmake file is localted with -DHPX_DIR=...?
<Abhishek09>
No hpx installation folder
<hkaiser>
can you read at all?
<hkaiser>
cmake told you: <quote>Add the installation prefix of "HPX" to CMAKE_PREFIX_PATH or set "HPX_DIR" to a directory containing one of the above files. </quote>
<hkaiser>
please read this: "set "HPX_DIR" to a directory containing one of the above files."