Package Details: anaconda 2019.03-1

Git Clone URL: (read-only)
Package Base: anaconda
Description: Completely free enterprise-ready Python distribution for large-scale data processing, predictive analytics, and scientific computing.
Upstream URL:
Licenses: custom
Submitter: flexiondotorg
Maintainer: petronny
Last Packager: petronny
Votes: 72
Popularity: 1.931422
First Submitted: 2013-06-24 14:51
Last Updated: 2019-04-15 04:56

Latest Comments

1 2 3 4 5 6 ... Next › Last »

chrisjbillington commented on 2019-01-23 17:41

I'm not using it with sudo, in fact I'm not using it at all, I'm just opinionated.

But I tested what I thought would break and it's not as bad as I thought.

Because cloning really does make basically a new installation, upgrading Anaconda via the AUR does not mess with the existing clone. Though it would still make me wary.

Also, making a clone is essentially a completely separate installation not tracked by pacman, one which will not be updated when the AUR package updates. So it doesn't break, but you're also not gaining anything from using a package manager. It's even in your home directory (I didn't realise this).

Ok, maybe it's not so bad having this as an AUR package. It is still pretty odd, but I can see it might be a good workflow for some.

petronny commented on 2019-01-23 06:02

@chrisjbillington Another one using this package with sudo.
You shouldn't do that and please create a clone if you want to modify the environment.

$ conda create -n myroot --clone /opt/anaconda
$ source activate myroot
$ conda install pyqt

chrisjbillington commented on 2019-01-22 15:48

I really think Anaconda ought not to be an AUR package. It is a package manager and environment in its own right, and will leave files that are untracked by pacman, leading to advice elsewhere in this thread to install with --force to upgrade, and to clone the conda environment before installing packages. Neither of these are good ideas in the long run, as after an update or two to this AUR package, both will eventually break with files existing within the same directory that are for incompatible with each other with respect to compiler, Python version, or other things. You should install Anaconda exactly once (whether via this package or not) and then not update it from an AUR package, updating only with conda itself. Personally I would just put it in my home folder by running the installer from pacman and the AUR add no extra value for this package, and it violates their assumptions of a package being a static set of files that can be upgraded.

mapcode.mind commented on 2018-11-03 11:55

it appears that anaconda is compiled w/o support for multiple processors. Is there a way for me to pass an argument like -j4 during install or anything. I have a 2-core 4-threads processor.

artemklevtsov commented on 2018-11-01 08:06 hash check failed.

petronny commented on 2018-06-11 17:16

@kbumsik @CaeZaR Build the package and install it with

$ pacman -U /path/to/anaconda.pkg.tar.xz --force

And you shouldn't use sudo conda to install things to /opt/anaconda. Create a clone and use it.

$ conda create -n myroot --clone /opt/anaconda
$ source activate myroot
$ conda install pyqt

CaeZaR commented on 2018-06-11 15:25

I have similar issues to @kbumsik when upgrading from 5.1.0 to 5.2.0.

All files report the following error:

anaconda: /opt/anaconda/XXX.XXX exists in filesystem

kbumsik commented on 2018-06-10 20:14

FYI, here is a desktop entry for anaconda-navigator, paste this to /usr/share/applications/anaconda-navigator.desktop:

[Desktop Entry]
Comment=Scientific PYthon Development EnviRonment - Python3
Exec=bash -c 'source /opt/anaconda/bin/activate && anaconda-navigator'

kbumsik commented on 2018-06-10 18:57

I cannot upgrade from 5.1.0 -> 5.2.0 because it says some files already exists. It looks like it conflicts with some packages I installed using conda. Is there a workaround?

jadelord commented on 2018-06-02 04:09

FYI, miniconda3 is now available in AUR, using a PKGBUILD nearly identical to the one used here.