We are currently exploring the installation of eScriptorium (as a first step just for internal use only) with the help of your magnificent documentation (https://github.com/UB-Mannheim/eScriptorium_Dokumentation/blob/main/Lokale_Installation_eScriptorium.md). I was able to get it running under Ubuntu 24.04 LTS (Noble Numbat) in a virtual machine, but had some initial struggles that i would like to share. I believe some changes should be made to the description of the installation. Either by providing a seperate section that describes changes specifically for Ubuntu 24.04 LTS in the affected steps or as a seperate page? Not quite sure which way you would prefer... maybe its even general enough to work from Ubuntu 20.04 upwards ... but i have no knowledge about WSL, so have not tested it there.
Also i cant really tell at the moment, if actually everything works as it should and if the vulnerabilites mentioned by NPM are actually harmful or not - the eScriptorium GIT progressed a lot and the interface changed a lot as well. I was able to segment, recognize and train a new model based upon an imported model (german_print/german_handwriting). So from this perspective everything so far worked out as it should.
Important: It is set that our servers will run with nothing less then the current stable LTS release, which is 24.04. We won't use older versions on new servers because of our policy.
Installation issues
Ubuntu 24.04 LTS ships with Python 3.12.x which causes the installation to fail, as we can't install and use necessary dependencies. This is because the virtual python environment will use the distros version, if not specifically provided another one.
e.g. Kraken (https://pypi.org/project/kraken/)
installation fails because --> Python <=3.11.99, >=3.8
So we need to install at least Python 3.8.x - the eScriptorium Wiki luckily mentioned this as well (https://gitlab.com/scripta/escriptorium/-/wikis/full-install). I had tested for Python 3.11.x to fullfil krakens dependency, but the installation failed at a later point, when checking via manage.py ... so even this version is to high.
So we would need (at least if not specifically) to install Python 3.8.x under Ubuntu 24.04 LTS. From what i have seen, only two steps would need to be changed then:
1. Installation relevanter Programme
This affects step 1, so the call would actually be:
# add custom ppa
sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt update
sudo apt install git postgresql postgresql-contrib libpq-dev redis-server gettext netcat-traditional jpegoptim pngcrush libvips build-essential python3.8 python3.8-dev python3.8-venv npm
Not sure if we actually require python3.8-pip as well? It's just mentioned in the Scripta-Documentation but not in yours ... so just as a mention.
7. Virtuelle Umgebung
Also we need to create a virtual python environment under this step with the now appropriate version:
python3.8 -m venv env
source env/bin/activate
pip install -U pip setuptools wheel
We can check the python version in our virtual environment simply by typing python --version ... in our case: Python 3.8.19
Everything else went through without any problems...
We are currently exploring the installation of eScriptorium (as a first step just for internal use only) with the help of your magnificent documentation (https://github.com/UB-Mannheim/eScriptorium_Dokumentation/blob/main/Lokale_Installation_eScriptorium.md). I was able to get it running under Ubuntu 24.04 LTS (Noble Numbat) in a virtual machine, but had some initial struggles that i would like to share. I believe some changes should be made to the description of the installation. Either by providing a seperate section that describes changes specifically for Ubuntu 24.04 LTS in the affected steps or as a seperate page? Not quite sure which way you would prefer... maybe its even general enough to work from Ubuntu 20.04 upwards ... but i have no knowledge about WSL, so have not tested it there.
Also i cant really tell at the moment, if actually everything works as it should and if the vulnerabilites mentioned by NPM are actually harmful or not - the eScriptorium GIT progressed a lot and the interface changed a lot as well. I was able to segment, recognize and train a new model based upon an imported model (german_print/german_handwriting). So from this perspective everything so far worked out as it should.
Important: It is set that our servers will run with nothing less then the current stable LTS release, which is 24.04. We won't use older versions on new servers because of our policy.
Installation issues
Ubuntu 24.04 LTS ships with Python 3.12.x which causes the installation to fail, as we can't install and use necessary dependencies. This is because the virtual python environment will use the distros version, if not specifically provided another one.
So we need to install at least Python 3.8.x - the eScriptorium Wiki luckily mentioned this as well (https://gitlab.com/scripta/escriptorium/-/wikis/full-install). I had tested for Python 3.11.x to fullfil krakens dependency, but the installation failed at a later point, when checking via manage.py ... so even this version is to high.
So we would need (at least if not specifically) to install Python 3.8.x under Ubuntu 24.04 LTS. From what i have seen, only two steps would need to be changed then:
1. Installation relevanter Programme
This affects step 1, so the call would actually be:
Not sure if we actually require
python3.8-pipas well? It's just mentioned in the Scripta-Documentation but not in yours ... so just as a mention.7. Virtuelle Umgebung
Also we need to create a virtual python environment under this step with the now appropriate version:
We can check the python version in our virtual environment simply by typing
python --version... in our case:Python 3.8.19Everything else went through without any problems...