Upgrading¶
While we try to make upgrades from one version of MDBenchmark to another as easy as possible, we are sometimes forced to break things along the way. This page gives some guidelines on how to migrate to specific versions of MDBenchmark, if you have been using a different version before.
To update from a previous MDBenchmark version to a new release, run the
following, if you are using conda
and your conda environment is called
benchmark
:
conda update -n benchmark mdbenchmark
and if you use pip
, run:
pip install mdbenchmark --upgrade
Refer to our installation guide, if you need help with conda/virtual environments.
Migrating to a newer version¶
Version 2.0.0¶
Note
TL;DR:
You need to uninstall the unneeded datreant.core
, datreant.data
and mdsynthesis
Python packages. Or just get rid of the current environment, e.g., conda env remove -n benchmark
if you are using conda
, and create it from scratch.
Starting from version 2.0.0, MDBenchmark migrated to a different dependency,
than it used before. While the specialized mdsynthesis
package was used in
previous version, we migrated to the more general datreant
package. This
leads to two things:
- Previously, every folder representing a specific number of nodes contained a
Sim.<uuid>.json
file, where<uuid>
is a UUID. This file contained all information representing a specific benchmark with its parameters. With the newer version ofdatreant
, the format was changed and now each of these folders contains a.datreant
folder instead. datreant
version 1.0 changed its package layout and led to the necessity of uninstalling previous packages. Because we were using a previousdatreant
version, you as a user will also need to uninstall those packages.
Uninstalling old packages¶
For proper migration, you need to uninstall the following three packages:
datreant.core
datreant.data
mdsynthesis
If you are using conda
and your environment is called benchmark
, run the
following command:
conda remove -n benchmark datreant.core datreant.data mdsynthesis
You can also decide to fully remove the environment (conda env remove -n
benchamrk
) and create it from scratch.
If you are using pip
simply run:
pip uninstall datreant.core datreant.data mdsynthesis