Why is conda trying to update packages using -no-update-dependencies?

Question

Often, when I try to install a new package, conda also wants to update other packages, even if I added the -no-update-dependencies switch. Updates seem "unnecessary" - as in most cases, only the last part of the version number has changed.

Today I wanted to install the mpld3 package, and conda wants to upgrade my python package from version 3.4.4-2 to 3.4.4-4, although I added the -no-update-dependencies switch.

How can I make conda to install mpld3 package without touching other packages?

C:\...>conda install -p pyenv --no-update-dependencies mpld3 Fetching package metadata: .... Solving package specifications: ........... Package plan for installation in environment C:\...\pyenv: The following packages will be downloaded: package | build ---------------------------|----------------- vs2010_runtime-10.00.40219.1| 0 1.1 MB python-3.4.4 | 4 31.7 MB mpld3-0.2 | py34_0 123 KB ------------------------------------------------------------ Total: 33.0 MB The following NEW packages will be INSTALLED: mpld3: 0.2-py34_0 vs2010_runtime: 10.00.40219.1-0 The following packages will be UPDATED: python: 3.4.4-2 --> 3.4.4-4 Proceed ([y]/n)? 

Answer 1

TL; DR;

Do not worry, your packages are not updated, only their build numbers, which should be harmless.

some explanation

The build number is updated, not the version of each installed package.

As you can see, it installs vs2010_runtime , this is due to updating vs2010_runtime , how its packages were created using functions. Since you have Python 3.4 (which is built using Visual Studio 2010), it installs the VS 2010 runtime. It will install the VS 2015 runtime if you install Python 3.5.

The numbers / lines of the assembly should not violate anyone, because they must be fixed during the assembly of this package (for example: you added a flag to the compilation that you were not going to).

It can be argued that conda should update packages (the same version, different build numbers) when --no-update-dependencies present because you may encounter an incorrect installation.

Answer 2

I want to say right away that these options are deprecated in new versions of conda. For example, in conda version 4.6.7 you should use "conda install --no-deps yourpackage"