From edfc583c58c8ac0a5deb6029de4b55d430d30386 Mon Sep 17 00:00:00 2001 From: anothersimulacrum <42699974+anothersimulacrum@users.noreply.github.com> Date: Sat, 19 Sep 2020 13:44:44 -0700 Subject: [PATCH] Minor building documentation updates Fix some typos, clarify some things, and TechWizzart has confirmed that the windows instructions work, so remove the disclaimer. --- BUILDING.md | 33 +++++++++++++++------------------ 1 file changed, 15 insertions(+), 18 deletions(-) diff --git a/BUILDING.md b/BUILDING.md index bd49a07..46b6246 100644 --- a/BUILDING.md +++ b/BUILDING.md @@ -16,27 +16,24 @@ python ./toy_model.py Below are more specific instructions to get setup for various operating systems. -## Windows (not tested) - -As stated, these are not tested, but these are the hypothetical steps to set up and build CLAuDE on Windows. -Please report if these do or do not work, so documentation can be updated. +## Windows ### 1. Clone or download the repository If you plan on contributing to this project, you will need to use git in some way to clone the project. Otherwise, simply selecting to download the repo will do. -Various git programs, such as [GitHub Desktop](https://docs.github.com/en/desktop/contributing-and-collaborating-using-github-desktop/cloning-a-repository-from-github-to-github-desktop), [GitKraken](https://support.gitkraken.com/working-with-repositories/open-clone-init/), and [git from Windows](https://gitforwindows.org/) are available to clone the repository with git. Each of these provide different levels of features and different levels of complexity. +Various git programs, such as [GitHub Desktop](https://docs.github.com/en/desktop/contributing-and-collaborating-using-github-desktop/cloning-a-repository-from-github-to-github-desktop), [GitKraken](https://support.gitkraken.com/working-with-repositories/open-clone-init/), and [git for Windows](https://gitforwindows.org/) are available to clone the repository with git. Each of these provide different levels of features and different levels of complexity. ### 2. Install Python, pip and a compiler To build this project, you will need Python, a compiler for the C code that Cython generates, and several python libraries that will be installed with pip. -Pip is a package manager, a program which simplifies installation of software. +Pip is a package manager, a program which simplifies installation of software, in this case, python libraries. -You can install python from [it's website](https://www.python.org/downloads/windows/). +You can install python from [its website](https://www.python.org/downloads/windows/). -If you do not already have pip installed, follow the instructions on the [pip website](https://pip.pypa.io/en/stable/installing/) to install it. +If you do not already have pip installed (it should be installed with python if you have an up to date version), follow the instructions on the [pip website](https://pip.pypa.io/en/stable/installing/) to install it. #### Compiler A compiler must then be installed. According to [the cython documentation](https://cython.readthedocs.io/en/latest/src/quickstart/install.html), you can use either [MinGW](https://osdn.net/projects/mingw/releases/) or Microsoft Visual C. If you choose to use MinGW, follow the instructions in the [cython documentation](https://cython.readthedocs.io/en/latest/src/tutorial/appendix.html) to get it set up. -If you choose to use Microsoft Visual C/C++, follow the instructions on the [Python website](https://wiki.python.org/moin/WindowsCompilers) +If you choose to use Microsoft Visual C/C++, follow the instructions on the [Python website](https://wiki.python.org/moin/WindowsCompilers). ### 3. Use pip to install python libraries Using command prompt, install the pip packages [cython](https://pypi.org/project/Cython/), [numpy](https://pypi.org/project/numpy/), and [matplotlib](https://pypi.org/project/matplotlib/), as well as [setuptools](https://pypi.org/project/setuptools/), if for some reason it is not already installed. @@ -54,7 +51,7 @@ python claude_setup.py build_ext --inplace This will convert the `.pyx` files in the repository to C files, then compile them. ### 5. Run the model -Simply run in your command prompt, while in the directory containing this file. +Simply run this in your command prompt, while in the directory containing this file. ``` python toy_model.py ``` @@ -70,7 +67,7 @@ The following programs and libraries are required to build CLAuDE: - numpy - matplotlib -If you're using apt as your package manager, these correspond to the packages `git` `python3` `python3-setuptools` `cython3` `python3-numpy` `python3-matplotlib` +If you're using apt as your package manager, these correspond to the packages `git` `python3` `python3-setuptools` `cython3` `python3-numpy` `python3-matplotlib`. ### 1. Clone the repository. Using the git command line, and cloning via https, this command will do it: @@ -103,17 +100,16 @@ This means that you did not install the equivalent of apt's `python3-dev` packag This likely means that the compiler building the libraries is not being told to search the proper directories for header files to include. One such case where this can happen is when you have libxcrypt installed, so Python.h includes ``, but setuptools does not tell the compiler where to search for that. -There is probably a proper solution to this, but I don't know it,so for now, you can work around this by telling it manually what to include. -To find the header you need to include, you can use `find`. +There is probably a proper solution to this, but I don't know it, so for now, you can work around this by telling it manually what to include. For example, with following error: ``` /usr/include/python3.8/Python.h:44:10: fatal error: crypt.h: No such file or directory 44 | #include | ``` -You need to find where the crypt.h header is to tell setuptools where to include it, so you can use find. -You can use find in the format `find dir/ -name "pattern"` to recursively search directories and get the full path of a the header file you are searching for. -If the #include has a directory included, use the directory which contains that directory. +You need to find where the crypt.h header is to tell setuptools where to include it, so you can use the aptly named `find`. +You can use `find` in the format `find dir/ -name "pattern"` to recursively search directories from `dir/` and get the full path of the header file you are searching for. +If the #include has a directory included (e.g. `#include `), use the directory which contains that directory (if `foo.h` is in `/usr/include/alice/blah/foo.h`, you want `/usr/include/alice`). So for example, you use `find /usr -name "crypt.h"` to find the full path of the crypt.h header that Python.h is looking to include. I got the output `/usr/include/tirpc/rpcsvc/crypt.h`, so now I know where that file is found, and I can instruct setuptools to include it. @@ -138,9 +134,10 @@ Please report if these do or do not work, so documentation can be updated. ### 1. Install HomeBrew HomeBrew is a package manager, software which simplifies the installation of various software. -You can install HomeBrew from [it's website](https://brew.sh/). +You can install HomeBrew with instructions from [its website](https://brew.sh/). ### 2. Clone the repository +If you plan on contributing to this project, you will need to use git in some way to clone the project. Otherwise, simply selecting to download the repo will do. Various git programs, such as [GitHub Desktop](https://docs.github.com/en/desktop/contributing-and-collaborating-using-github-desktop/cloning-a-repository-from-github-to-github-desktop), [GitKraken](https://support.gitkraken.com/working-with-repositories/open-clone-init/), or just git are available to clone the repository. Each of these provide different levels of features and different levels of complexity. git can be installed through homebrew using `brew install git`. Using git, you can clone the repository via https with this command: @@ -159,7 +156,7 @@ Then, use `pip`, a python package manager that is installed with python to insta ``` pip install cython pip install numpy -pip install matplot lib +pip install matplotlib ``` ### 4. Using Cython, compile `claude_low_level_library` and `claude_top_level_library`