We are currently in the process of evaluating the deployment possibilities for PyRevit on a company scale, encompassing approximately 300-400 individuals.As we embark on this initiative, we seek to ensure that our deployment strategy is effective and optimized.
Outlined below is our proposed plan:
1- First , We intend to create a forked repository from the main PyRevit repo.adjustments will be made to bundle and YAML files, with the objective of concealing specific buttons, such as the destructive wipe/purge functionality, to tailor the user experience
2 - This repo will be hosted on a shared local server within the company. where all users will be able to access it. The management of updates will be done through pull requests
3 - The deployment of PyRevit core through package installer for every user, including .ini file to custom settings and and paths to other extensions currently under development. This will be done through a script to manage silent a setup
Anyone has recommendations for alternative methods or workflows that could potentially optimize our deployment strategy further?
You could also host your pyrevit version and extensions directly on github and drive the install and update of extension through the pyrevit extend ui command
I suggest you use the develop-4 branch to get the latest changes. There is a small but major bugs in the latest release regarding the lib2git library that could forbid you using CLI onto github repos
I’ve got pyRevit and our toolbar on about 400 machines where I work. Easiest way we found was actually using Microsoft Intune and the standard installer vs trying to get all users working via git setups. We then have a separate intune installer to write custom toolbars as folders to %APPDATA%/extensionName, then we update the pyrevit ini in %APPDATA%/pyRevit with the extension folders path to ensure all users get the path to the custom toolbar as well.
An app like PDQ could achieve this as well. I’m guessing at your company scale the IT department has some means of adding/replacing folders on machines, which at the end of the day is moreorless what pyRevit extensions are.
I know git/CLI is the intended automatic way to achieve this, but I suspect you may also find limitations when it comes to getting the average user to work this way or getting IT comfortable using this approach vs more typical distribution methods other apps use already.
One word of warning is try to get updates out to users when they’re not mid session ideally. So queue up before log out or on log in before they get time to open Revit. I had a change get pushed out in the day and it had an event interceptor for sync timing which cancelled the sync event due to partial deployments on some machines. Not fun…!
