ip-tools | enables calculations that involve ip addresses | TCP library

To the best of my knowledge (which is several years of using pip-tools), pip-tools will always give you a stable tree so long as you do then install dependencies only from the "locked" requirements file.

Unless you've deeply internalized the format, you can seldom look at the dotted quad notation and have any idea what the subnetting is for anything other than a multiple of 8.

The binary representation of those networks is:

I got the fix finally \o/

So now my predicament is: should I use --allow-unsafe? What are the implications of doing this?

Yes, you should. This option allows you to pin in requirements.txt the following packages: distribute, pip, and setuptools. If you don't care, go ahead!

But why is this unsafe?

AFAIK, those packages could be considered unsafe for the following reasons:

• Changing the setuptools may cause conflicts with pip ( distribute is the legacy wrapper of setuptools, and it's deprecated since 2013).
• Changing pip could break pip-tools itself or your system pip.

The --allow-unsafe option most likely will be deprecated in the near future, see discussions in pip-tools' and pip's issue-trackers.

From the zappa tail output, it seems that your function tries to connect localhost but since it is lambda I doubt there is a running postgres instance on localhost. You should update your connection settings for postgres (correct address of the remote postgres instance). If you are using sqlalchemy, you can check if SQLALCHEMY_DATABASE_URI is correct.

That happens because you are not installing it globally (system wide). For it to be available in your path you need to install it using sudo, like this:

So with the details as provide in the question, we have the scenario where a Python script and the environment it will be executed at has:

1. Dependency on some packages;
2. No virtualenv is used, but the Python user install directory (as the ec2-user user) is used instead for simplicity for the installation of the script's dependencies;
3. The script however requires root privileges to run (due to the usage of low ports, with the elevated permission achieved via sudo (as the root user).

Naturally, running a script as root while the dependencies were installed local to a user other than root will mean that none of the dependencies will actually be accessible by that script. As the dependencies are installed under the site.USER_BASE for the ec2-user, in order for the script to be able to import its dependency while being executed under the root user, the location may be defined using the PYTHONUSERBASE environment variable.

To achieve that while using sudo, try:

I had the exact same problem as you. 3 bytes were different near the tail end of the .so that was packed in the .apk (app/build/outputs/apk/debug). The three bytes changed as follows:

• 0x00 to 0x04
• 0x04 to 0x00
• 0x0a to 0x14 (the difference is the same as your byte changing from 0x08 to 0x12)

I was digging around in the app/build folder and I saw that the .so file was also located in intermediates/transforms/mergeJniLibs and intermediates/transforms/stripDebugSymbol. The md5sum of the .so file in mergeJniLibs matched the original .so file's sum. The md5sum of the .so file in stripDebugSymbol was the one that was different. Further investigation led me to add this to the android section of Module's build.gradle (see also: PackagingOptions doNotStrip documentation ):

The document I saw is outdated. The workable version is this