I'm following the instructions for Building From Source at https://pillow.readthedocs.io/en/stable/installation.html#building-from-source for building on Debian Linux. Pillow seems to work fine on my system after I follow these steps, but it doesn't seem to be really building from source.
$ git clone https://github.com/python-pillow/Pillow.git
Cloning into 'Pillow'...
remote: Enumerating objects: 67869, done.
remote: Counting objects: 100% (163/163), done.
remote: Compressing objects: 100% (95/95), done.
remote: Total 67869 (delta 90), reused 128 (delta 68), pack-reused 67706
Receiving objects: 100% (67869/67869), 83.17 MiB | 9.39 MiB/s, done.
Resolving deltas: 100% (50202/50202), done.
$ sudo apt-get install python3-dev python3-setuptools
[sudo] password for dan:
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
python3-dev is already the newest version (3.9.2-3).
python3-setuptools is already the newest version (52.0.0-4).
0 upgraded, 0 newly installed, 0 to remove and 8 not upgraded.
Remove all the source files from libImaging directory to prove its not building from source:
$ cd Pillow
$ rm ./src/libImaging/*
Continue to follow the Building From Source instructions at https://pillow.readthedocs.io/en/stable/installation.html#building-from-source for Debian Linux:
$ python3 -m pip install --upgrade pip
Defaulting to user installation because normal site-packages is not writeable
Requirement already satisfied: pip in /home/dan/.local/lib/python3.9/site-packages (22.3.1)
$ python3 -m pip install --upgrade Pillow --no-binary :all:
Defaulting to user installation because normal site-packages is not writeable
DEPRECATION: --no-binary currently disables reading from the cache of locally built wheels. In the future --no-binary will not influence the wheel cache. pip 23.1 will enforce this behaviour change. A possible replacement is to use the --no-cache-dir option. You can use the flag --use-feature=no-binary-enable-wheel-cache to test the upcoming behaviour. Discussion can be found at https://github.com/pypa/pip/issues/11453
Collecting Pillow
Using cached Pillow-9.3.0.tar.gz (50.4 MB)
Preparing metadata (setup.py) ... done
Installing collected packages: Pillow
DEPRECATION: Pillow is being installed using the legacy 'setup.py install' method, because the '--no-binary' option was enabled for it and this currently disables local wheel building for projects that don't have a 'pyproject.toml' file. pip 23.1 will enforce this behaviour change. A possible replacement is to enable the '--use-pep517' option. Discussion can be found at https://github.com/pypa/pip/issues/11451
Running setup.py install for Pillow ... done
Successfully installed Pillow-9.3.0
Pillow works on my system, but as I removed a large number of source files, it doesn't appear that I'm really building from source. The output says "Using cached Pillow-9.3.0.tar.gz" which implies it isn't using what I cloned. I think I'm following the all directions. Did I do something wrong or are the directions incomplete?
I built the latest Pillow from source inside a dockerized debian:latest like this:
# Start debian:latest docker
docker run -it debian:latest
# Install all the junk you need and upgrade pip
apt update && apt install git python3-dev python3-setuptools python3-pip cmake libjpeg-dev
python3 -m pip install --upgrade pip # this line MAY not be strictly necessary
# Clone PIL/Pillow from Github
git clone https://github.com/python-pillow/Pillow.git
# Install from source
cd Pillow
python3 -m pip install .
Related
I have two servers that are running CentOS 7.8 with all the latest patches as of the date of this post. On one server I can create a venv, and on the other I get an error. Here is the command used to create the venv:
python3.6 -m venv venv
The error is:
Error: Command '['/root/venv/bin/python3.6', '-Im', 'ensurepip', '--upgrade', '--default-pip']' returned non-zero exit status 1.
So it seems there is a problem after the venv is created. In fact, it makes the links to the python interpreters in venv/bin, but the lib64/python3.6 and lib/python3.6 directories that are created have only empty site-packages in them. I get the problem whether running the command with root privileges (as above) or as a non-privileged user. On the other system, which is not quite the same hardware but very similar, I get no errors.
I have tried a bunch of things including completely uninstalling all the python3 rpms and reinstalling them on both systems. The python3 modules installed are:
python3, python3-devel, python3-libs, python3-other-rpm-macros, python3-pip, python3-rpm-generators, python3-setuptools, python3-tkinter, python3-wheel. All these are installed using the centos repos.
I have searched the internet and one thing that works is to use --no-pip. However, not surprisingly, if I do that I get a venv with no pip, which does not work very well. The version of pip that is installed with the rpm is version 9.0.3. However, I have tried upgrading pip:
python3.6 -m pip install --upgrade pip
This installs pip 20.1. Interestingly, it is installed in /usr/local/lib/python/site-packages. In any event, it doesn't change anything and I get essentially the same error message.
I have even gone as far as rsyncing the /usr/lib/python3.6 and /usr/lib64/python3.6 (using -avu --delete) on the two servers so that they are identical. All to no avail. I am at a loss to know what to do.
I can of course hack things to make them work, but since I am using this to automatically build test environments I really need to get it working automatically. Suggestions would be greatly appreciated.
I've had the same issue on CentOS Linux release 7.9.2009 (Core) and solved it by installing the following system packages:
yum install python36-devel
yum install python36-setuptools
yum install python36-virtualenv
python3.6 -m pip install --upgrade pip
Used virtualenv to install the venv into the project-root:
python3 -m virtualenv venv-dev
After sourcing the venv everything looks good and I could install the needed requirements:
. ./venv-dev/bin/activate
python3 --version
Python 3.6.8
pip3 --version
pip 21.1.1 from /project/root/dir/venv-dev/lib/python3.6/site-packages/pip (python 3.6)
pip3 install -r requirements/dev.txt
Successfully installed ...
I installed Scipy using the command sudo apt-get install python-numpy python-scipy and looks like the latest version of both are installed. But when I try to install a python package named wfdb using pip install wfdb, it get an error while Building wheel for scipy (PEP 517).
ERROR: Failed building wheel for scipy
Running setup.py clean for scipy
ERROR: Command errored out with exit status 1:
command: /usr/bin/python -u -c 'import sys, setuptools, tokenize; sys.argv[0] =
'"'"'/tmp/pip-install-
fhfmfpqi/scipy/setup.py'"'"'; __file__='"'"'/tmp/pip-install-
fhfmfpqi/scipy/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)
(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code,
__file__,
'"'"'exec'"'"'))' clean --all
cwd: /tmp/pip-install-fhfmfpqi/scipy
Complete output (9 lines):
`setup.py clean` is not supported, use one of the following instead:
- `git clean -xdf` (cleans all files)
- `git clean -Xdf` (cleans all versioned files, doesn't touch
files that aren't checked into the git repo)
Add `--force` to your command to use it anyway if you must (unsupported).
----------------------------------------
ERROR: Failed cleaning build dir for scipy
Failed to build scipy
ERROR: Could not build wheels for scipy which use PEP 517 and cannot be installed directly`
I upgraded pip using pip install --upgrade pip.
python version used: 3.7
Can I get some help on this?
Clean out the install (I don't know which system you're using) i.e. get rid of the $PATH variables and just do a fresh install.
If you're using Anaconda:
conda install -c anaconda scipy
should do it
also the goops old
pip install scipy
gets it done as well. May I ask why you approached your installation in this way (again, I don't know which system you're on)
Never do sudo pip. Instead do pip install --user. Or (better) work in a virtualenv.
EDIT: typo: there is no space between -- and user.
I am running ubuntu 16.04 and have some python code that uses features from python 3.6. I have been able to install some libraries (scipy) with python3.6 and pip3.6. However there have been a few libraries (pycurl) for which using pip install is not working :
$sudo python3.6 -m pip install pycurl
Requirement already satisfied: pycurl in /usr/lib/python3/dist-packages (7.43.0)
This does not allow me to import pycurl in my python3.6 environment.
And if I instead use upgrade I get an error about pycurl being a distutils installed package.
How can I get access to pycurl for my python3.6 install?
$ sudo python3.6 -m pip install --upgrade pycurl
The directory '/home/riaps/.cache/pip/http' or its parent directory is not owned by the current user and the cache has been disabled. Please check the permissions and owner of that directory. If executing pip with sudo, you may want sudo's -H flag.
The directory '/home/riaps/.cache/pip' or its parent directory is not owned by the current user and caching wheels has been disabled. check the permissions and owner of that directory. If executing pip with sudo, you may want sudo's -H flag.
Collecting pycurl
Downloading https://files.pythonhosted.org/packages/77/d9/d272b38e6e25d2686e22f6058820298dadead69340b1c57ff84c87ef81f0/pycurl-7.43.0.1.tar.gz (195kB)
100% |████████████████████████████████| 204kB 2.8MB/s
Installing collected packages: pycurl
Found existing installation: pycurl 7.43.0
Cannot uninstall 'pycurl'. It is a distutils installed project and thus we cannot accurately determine which files belong to it which would lead to only a partial uninstall.
So the problem seems to have been that it was installed in /usr/lib/python3/dist-packages. By running sudo pip3 install --upgrade pycurl It was moved to : /usr/local/lib/python3.5/dist-packages and then sudo python3.6 -m pip install pycurl could install.
I don't know why this was a problem, but it resolved the issue.
I have a CentOS 7 (EPEL-Repo is registered) but I do not find the package for CPP-REST (libcpprest-dev in Debian) in the repos.
Do I need to install it from vanilla or is there a package in CentOS?
Thanks!
Short answer...
There is not an "official" CentOS 7 package at this time (2018-10-03) for cpprestsdk (libcpprest-dev in Debian).
Long answer (sharing this in case it helps anyone)...
You could build cpprestsdk from source, but the instructions "How to build for Linux" (https://github.com/Microsoft/cpprestsdk/wiki/How-to-build-for-Linux) do not include CentOS or RHEL (at the time of this writing, 2018-10-03).
Before building cpprestsdk on CentOS 7, you'll need to update boost (at time of this writing CentOS 7 repos have boost 1.53, but 1.54 is required, and 1.68 is latest). You can do this to update (using boost 1.68 as example):
cd && wget https://dl.bintray.com/boostorg/release/1.68.0/source/boost_1_68_0.tar.gz
tar -xzvf boost_1_68_0.tar.gz
cd ~/boost_1_68_0
./bootstrap.sh --prefix=/opt/boost
sudo ./b2 install --prefix=/opt/boost --with=all
After updating boost, I was able to build cpprestsdk on CentOS 7 like this...
Install some needed tools/libraries:
sudo yum -y install git cmake3 openssl-devel gcc-c++ make
Download the cpprestsdk project:
git clone https://github.com/Microsoft/cpprestsdk.git casablanca
cpprestsdk wants to treat warnings as errors. I had to turn this off by commenting out this line in ~/casablanca/Release/CMakeLists.txt:
#set(WERROR ON CACHE BOOL "Treat Warnings as Errors.")
Prepare/build cpprestsdk project:
cd casablanca/Release
mkdir build.release && cd build.release
cmake3 .. -DCMAKE_BUILD_TYPE=Release -DBOOST_ROOT=/opt/boost
make
Install header files and library
sudo make install
Notice that I am calling cmake3 (not cmake) and I am specifying the boost root; which differs from the official instructions that appear to be written and tested for Ubuntu (not CentOS or RHEL). To build my projects using cpprestsdk, I had to use ldconfig to find the library paths.
How do I install PIL?
>pip install PIL
Downloading/unpacking PIL
Could not find any downloads that satisfy the requirement PIL
Some externally hosted files were ignored (use --allow-external PIL to allow).
Cleaning up...
No distributions at all found for PIL
Storing debug log for failure in /root/.pip/pip.log
>pip uninstall PIL
Can't uninstall 'PIL'. No files were found to uninstall.
pip install PIL --allow-external PIL --allow-unverified PIL
This is due to changes in the new version of Pip. Run pip --version and I'm willing to bet you are running 1.5. See the changelog here. This new default behavior enhances security. In PIL's case, the file you are installing actually comes from effbot.org (thus --allow-external) and PyPi doesn't have a checksum to guarantee validity (thus --allow-unverified).
Also, you might consider using the Pillow replacement to PIL.
Updated info for those reading in 2016:
--allow-external
and
--allow-unverified
were recently deprecated. Installing packages external to PyPi using pip is no longer supported: http://www.python.org/dev/peps/pep-0470/
As an alternative, when you really need to install that external package, you can download the source code and run its setup.py. For example, for PIL 1.1.7, download from http://www.pythonware.com/products/pil/, then:
$ tar xvfz Imaging-1.1.7.tar.gz
$ cd Imaging-1.1.7
$ python setup.py install
(^ from the PIL README)
If you only want to install the package to a specific virtualenv, you can just activate your virtualenv first. ** thanks #Caumons
Alternatively, substitute the path to your virtualenv for 'python' in the third line, e.g.:
$ /home/username/virtualenv-name/bin/python setup.py install