New ask Hacker News story: Ask HN: Do we need a better workflow solution exist, or is it already out there?

Ask HN: Do we need a better workflow solution exist, or is it already out there?
2 by mattewong | 1 comments on Hacker News.
I've been on the hunt for some time for a workflow solution that: 1. bifurcates administrator vs user experiences. So for example, an admin defines a universe of possible actions, and a user defines a workflow consisting of one or more triggers that, when fired, executes some series of those predefined actions in the flow that the user defines 2. is no-code. If the admin must sometimes work with a config file (such as JSON), that's OK. But none of this python or coding business-- not that I have anything against that myself, but it's just the use case we need to address. For non-admin users, it should have a graphical UI 3. is versatile. For example, I'd like to be able to define an action to invoke any executable I choose, or to make any REST call I define 4. can dynamically chain each step's output to any of the next steps' input. For example, let's say I'm a regular user, and my admin has defined 2 actions I can use: XLSX_to_CSV, which converts X.xlsx to output/X.csv, for any X, and CSV_to_SQLITE, which converts XX.csv to output/XX.db, for any XX. Then I should be able to define a new workflow, using a UI, with a trigger that, for example, will monitor folder X that I specify, and upon receiving an XLSX file named Y.xlsx, will execute action 1 followed by action 2, to produce output/Y.csv and output/Y.db. This should all sound pretty basic, because it is. I've looked at Airflow, Argo, OpenWDL and others, as well as explored whether the requirements could be met with a collection of technologies (such as something plus WuFoo together)-- but so far, they all seem to fall short. The closest thing I have seen is https://ift.tt/zHoArmh, but that does not let an admin user define actions for regular users to then mix and match (but admittedly, it is otherwise very close to what I'm looking for). Any suggestions on other solutions that may fit the bill? Or alternatively, if not, how might I go about gauging whether there might be sufficient interest to opensource a project for doing so (a decent amount of which is already built, since I needed some solution and built a workable one already, though it could use a facelift)?

Comments