Where communities thrive

  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
    Adam Scislowicz
    @wolfv I do use the Threads::Threads method in the library link section after PRIVATE, after earlier using the find_package mandatory technique to get that name populated
    Wolf Vollprecht
    please check the outptu with -DCMAKE_VERBOSE_MAKEFILE=ON and check if it's properly adding -pthread?
    Adam Scislowicz
    @wolfv ah, thank you, I will do that. I was also going to compare the env, between my pip run and my conda mambabuild run (or conda build) run but I was leaning towards outsourcing this as I'm mostly zoomed in on another aspect of the program right now. Although I do intend to build more conda packages in the future.
    Wolf Vollprecht
    well if you are interested in contracting with us, you can hit us up (info@quantstack.net)
    Adam Scislowicz
    yes. I'll send you an email right now

    Hi all, I'm trying to use micromamba inside a Dockerfile and am having difficulty. The file:

    FROM ubuntu:18.04
    RUN apt-get update && apt-get install --yes wget
    RUN wget -q -O - https://micro.mamba.pm/api/micromamba/linux-64/latest | tar -xvj bin/micromamba && \
        ./bin/micromamba shell init --shell bash --prefix ~/micromamba && \
        echo "micromamba activate" >> ~/.bashrc
    RUN micromamba install --name base --yes numpy -c conda-forge

    and the result of building:

    Step 4/4 : RUN micromamba install --name base --yes numpy -c conda-forge
     ---> Running in 792a0331ab4c
              __  ______ ___  ____ _____ ___  / /_  ____ _
             / / / / __ `__ \/ __ `/ __ `__ \/ __ \/ __ `/
            / /_/ / / / / / / /_/ / / / / / / /_/ / /_/ /
           / .___/_/ /_/ /_/\__,_/_/ /_/ /_/_.___/\__,_/
    WARNING 'root_prefix' set with default value: /root/micromamba
    ERROR   Could not use default 'root_prefix': /root/micromamba
    ERROR   Directory exists, is not empty and not a conda prefix.
    The command '/bin/sh -c micromamba install --name base --yes numpy -c conda-forge' returned a non-zero code: 1
    From googling around (here, here, here, and here), I gather that the issue is likely related to how the final docker command is run. I have tried a number of iterations based on what I've read but with no success.
    I REALLY like the idea of micromamba but using it to set up a single environment container has been pretty painful. I just want to install some packages in the base environment so they are there when I need them. Am I going about this the wrong way? Thanks for your help :)
    Follow up: I was able to do what I wanted using the mambaorg/micromamba image from dockerhub but I stand by my original question. What was I doing wrong then? Also, please let me know if this is the wrong venue, thanks!
    Sylvain Corlay
    you are downloading micromamba in /root/ and are trying to set the root prefix to the same path.
    1 reply
    hi again @wolfv ! Does the virtual host for voila have to be set up on the root of the server for everything to function correctly? Like getting the correct path for the static files, for example?
    Anton Kulaga
    Guys, are there any guides how to activate micromamba and use executables from environment inside a bash script?
    Hi all. I am looking for articles about jupyter hub in the enterprise that can be used to put together in a map for higher up management who is not vwry technical. Its about making them to understand the value of jupyter hub in a data driven organization. Somehow i find so little in the web. Any suggesrion where to look ? Or any concrete succesfull stories around using jupiter hub in the enterpeise? Thanks.
    Is there a mamba equivalent of conda run?
    Jason Grout
    Dave, that looks like a great list. Perhaps we should have something like that on jupyter.org, perhaps a Jupyter in the News section?
    Dave Hirschfeld
    I've also had to argue the case for Jupyter to management so I've kept a list of some references that might be considered more authoritative or tell a story about why it's a useful tool for data science / analytics
    Dave Hirschfeld
    I think it's useful for others in such a situation so it might be good to keep a list of references/publications on the official Jupyter site
    Ray Bell
    Is there a way of clearing the cache when doing mamba env update? Sometimes I don't get the latest version of a package and i'm guessing and older is cached and it picks that up?
    you can do a mamba clean --all
    Wolf Vollprecht
    There might also be another package holding the other one back
    Christian Roth
    @wolfv, do I see correctly that mamba remove --force is not yet implemented? Unlike conda, for me it does try to remove all dependent packages as well.
    Wolf Vollprecht
    yep, it's not
    it's correctly implemented for micromamba
    we just need to copy the implementation to mamba
    Tim Paine
    whats a good way to check a dep set with mamba (e.g. from a conda forge package)
    7 replies
    Bhavyansh Mishra
    Hello everyone. I am currently trying out xtensor for the first time, and trying to load it as a submodule in CMake. Is that a decent approach to using xtensor on Ubuntu 20.04? I've been getting multiple compilation errors regarding xtl.
    /usr/include/xtl/xvariant_impl.hpp:1287:35: error: wrong number of template arguments (1, should be at least 2)
     1287 |               case B + 29: return MPARK_DISPATCH(B + 29);
          |                                   ^~~~~~~~~~~~~~
    /usr/include/xtl/xvariant_impl.hpp:1198:16: note: provided for ‘template<bool B, class R, class ... ITs> struct mpark::detail::visitation::base::dispatcher1198 |         struct dispatcher;
          |                ^~~~~~~~~~
    /usr/include/xtl/xvariant_impl.hpp:1287:35: error: expected ‘;’ before ‘)’ token
     1287 |               case B + 29: return MPARK_DISPATCH(B + 29);
          |                                   ^~~~~~~~~~~~~~
    /usr/include/xtl/xvariant_impl.hpp:1287:35: error: expected primary-expression before ‘)’ token
    /usr/include/xtl/xvariant_impl.hpp:1288:35: error: wrong number of template arguments (1, should be at least 2)
     1288 |               case B + 30: return MPARK_DISPATCH(B + 30);
          |                                   ^~~~~~~~~~~~~~
    /usr/include/xtl/xvariant_impl.hpp:1198:16: note: provided for ‘template<bool B, class R, class ... ITs> struct mpark::detail::visitation::base::dispatcher1198 |         struct dispatcher;
          |                ^~~~~~~~~~
    /usr/include/xtl/xvariant_impl.hpp:1288:35: error: expected ‘;’ before ‘)’ token
     1288 |               case B + 30: return MPARK_DISPATCH(B + 30);
    Please advice. I'd highly appreciate all comments and suggestions.
    Johan Mabille
    Hello, we highly discourage the usage of git submodule, they are conisdered as bad practice regarding management of dependencies. A better approach is to either install the dependencies from sources as explained in the README of the different projects, or use a package manager. We provide packages of the xtensor-stack for mamba, and xtensor has also been packaged for debian. Packages for Ubuntu should be available soon
    Mirco Marahrens
    Hello, is there a minimal cmake example for the section on "Designing language bindings with xtensor" (https://xtensor.readthedocs.io/en/latest/bindings.html) somewhere available? I try to follow the procedure but failing on the last part on exposing the function to Python. Within C++ the generalization is working like charm (at least my tests are running successfully).
    3 replies
    Sylvain Corlay
    There isn't such a cmake file unfortunately.
    It is very dependent of the language btw, especially for e.g. R and Julia that have weird toolchains on windows.
    Patrick Bos
    hey, apologies for the lack of interaction on xtensor-fftw here! gitter isn't really in my system :) I see you were pinging me a few times @JohanMabille @pesterie, best to just hit me up in a github issue (or ask me to come on gitter via github :D chat can be more efficient and i'm always happy to help)
    hope you're all doing well! excited to see you're thinking about moving to c++17 btw, that should allow to clean up a lot of code with constexpr if
    Frank Schlimbach
    I am currently evaluating xtensor. I wonder which projects/products are using xtensor. Is there any information around that which you could share?
    Wolf Vollprecht
    @fschlimb I don't think we have such a list at this point
    You could use github code search to find a couple of users: https://github.com/search?q=xtensor.hpp&type=Code
    Mario Emmenlauer
    @fschlimb It may also help for which use case you are evaluating.
    Thorsten Beier
    @mircomarahrens you can try https://github.com/DerThorsten/cpp_cookiecutter . this is bit outdated but I personally use it still very often to generate python bindings for c++ projects with a CMake build system + xtensor
    Surbhi Sharma
    Hey folks! I needed some suggestion on a problem that I am facing and was redirected here for asking any questions regarding it. I have been trying to access multiple voila apps through nginx reverse proxy I also have a flask app which I am also using through the same reverse proxy. So, when I am doing so, if i redirected the location / directly to voila apps while voila app is running locally on my machine, it works. But when I am redirecting it for location /notebook1, I am not able to access the app. I the browser it shows it is loading and kernel is started but gets a 404 in accessing some static files ( ). I am a little new to nginx as well as voila. Any advice or suggestion regarding this would be of great help for me! Here is the nginx.conf file (https://pastebin.com/pHy6pgsU), and here is the voila.json file ( https://pastebin.com/gfx0gE8H), for now have been testing it just for one notebook if it works i will just add for the other two notebooks too with different ports. Here are the error message log I get when I try to open the voila through nginx reverse proxy (https://pastebin.com/C0h7f61S)
    Please let me know if any more info is needed.
    Duc Trung Le
    in your case you can try voila ... --base_url=/notebook1/
    Surbhi Sharma
    Thanks @trungleduc , it worked! \o/
    I set the base_url and also change the nginx.conf to redirect to updated url for voila app, and it worked. :D
    Tom de Geus
    Hi ! Is there a way to get the indices resulting from unique ?
    Benoit Bovy
    Hello! Quick question: is there a way to vectorize a C++ function (and expose it in Python as a Numpy ufunc) with arguments of non-POD type (e.g., a custom C++ class exposed in Python)? It doesn’t seem to be supported with pybind11… so I guess it’s not supported either by xtensor-python’s pyvectorize? I was wondering if there could still be a way around it, e.g., using a lightweight wrapper calling xtensor’s xvectorize?
    Johan Mabille
    Hello, I think it should work, pyvectorize does not rely on the vectorization mechanism from pybind11
    not 100% sure but you should definitely give it a try