Re: Refactoring and Optional Python Frontend for Isar Installer

["'Jan Kiszka' via isar-users" <isar-users@googlegroups.com>]

On 15.12.25 10:58, Kasturi S wrote:
> Dear Isar Community,
> 
> We are currently evaluating enhancements to the Isar installer and would
> like to share our plans, collect feedback, and align with upstream
> expectations before we proceed.
> 
> During our work with the existing installer (deploy-image-wic.sh), we
> observed that both the backend logic and the frontend elements (dialogs,
> progress bars, UI prompts) are tightly coupled within a single script.
> This makes it difficult to extend the installer with new capabilities or
> integrate more advanced user-facing features.
> 
>  
> 
> 1. Goals for Enhancing the Installer
> 
> We would like to upgrade the installer across the following areas:
> 
> •Advanced installation features
> 
> Support for RAID creation from menu, etc.
> 
> •Improved user experience
> 
> The current shell-based UI is functional but limited. We would like to
> explore a more modern and user-friendly installer experience.
> 
>  
> 
> 2. Proposed Refactoring
> 
> To enable better extensibility, we propose to separate the frontend and
> backend:
> 
> •Backend:
> 
> Extract all installation logic into reusable shell-based API functions,
> allowing a clean and consistent interface.
> 
> •Frontend:
> 
> Move user interaction elements (dialogs, text UI, progress
> visualization) into separate files to avoid mixing UI with logic.
> 
> This separation would keep the codebase more maintainable and make it
> easier to introduce new features or alternative frontends.
> 
>  
> 
> 3. Optional Python-Based Frontend
> 
> We are also considering offering an enhanced UI frontend using a Python-
> based framework. A Python UI would allow us to provide a richer, more
> intuitive interface that is difficult to achieve with pure shell scripting.
> 
> Our intention is:
> 
> •If upstream is open to a Python-based UI, we would like to contribute
> this as an optional frontend.
> 
> •If upstream prefers to keep the installer shell-based, we are happy to
> maintain the Python UI downstream while keeping the core backend APIs in
> upstream shell scripts.
> 
> In this model, the backend APIs would remain shell-based and stable,
> while both shell and Python frontends would call into the same
> underlying API functions. This would ensure compatibility and avoid
> duplicating backend logic.
> 
> Here is the basic example:
> 
> Shell api:
> 
> ```
> 
> sys_is_device_empty() {
> 
>     local -A ARGS
> 
>     local required=(device)
> 
>     api_args ARGS required[@] "$@" || {
> 
>         pack_return_data error "$_args_error" retval "1"
> 
>         return 1
> 
>     }
> 
>  
> 
>     local fn="${FUNCNAME[0]}"
> 
>     local device="${ARGS[device]}"
> 
>     log_info "$fn" "Checking if device '$device' is empty"
> 
>  
> 
>     if cmp /dev/zero "$device" -n 1M >/dev/null 2>&1; then
> 
>         pack_return_data error "" retval "0" status "empty"
> 
>     else
> 
>         pack_return_data error "" retval "1" status "not_empty"
> 
>     fi
> 
> }
> 
> ```
> 
> Calling api from shell :       
> 
> ```
> 
> #  Check if device is empty
> 
> echo "Checking if device /dev/sdb is empty..."
> 
> output=$(sys_is_device_empty device="/dev/sdb")
> 
> echo "Output: $output"
> 
> echo                                                                                           
> 
> ```
> 
> calling same api from python:
> 
> ```
> 
> import subprocess
> 
> def main():                   
> 
>    # Check empty status
> 
>     result = is_device_empty(device="/dev/sdb")
> 
>     print("Result:", json.dumps(result, indent=2))
> 
> ```
> 
> 4. Request for Feedback
> 
> Before we move ahead, we would appreciate feedback from the community on:
> 
> 1.The idea of splitting the installer into modular backend and frontend
> components
> 
> 2.Contributing backend installer logic as structured shell APIs
> 

This alone would definitely be welcome. It is a bit hard to find a good
abstraction when you only have one user in hands, but as you already
have a second one in mind (or already working?), it would be a great
chance to split backend and frontend.

> 3.The possibility of supporting an optional Python-based UI frontend
> 

At least I would not mind having that second option in-tree as well.
Would both help to keep the abstraction working and may be useful for
other real use cases as well.

> 4.Any concerns or suggestions regarding maintainability or integration
> 
> We want to ensure that our work aligns with the long-term vision of the
> Isar project and benefits both upstream and downstream users.
> 
> We look forward to the community’s thoughts and guidance on the best way
> to proceed.
> 

The original vision of pushing the installer to upstream was to have a
foundation for related work available, ideally in a form that more
complex, customized, branded etc. installer can build on top. In that
sense your ideas fit well. Make sure to present the code changes in
steps, likely similar as you described above, and the rest should just
be about details.

Jan

-- 
Siemens AG, Foundational Technologies
Linux Expert Center

-- 
You received this message because you are subscribed to the Google Groups "isar-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isar-users+unsubscribe@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/isar-users/0b7503b2-96d8-40e0-b9cb-d025c2f5abf3%40siemens.com.