Thanks for your detailed reply. I am coming more from a background of finding solutions to properly bill SaaS offerings to customers. In that sense, the initial requirements are way more simple, i.e. it would suffice to create a subscription or a contract and an invoice via the REST API and provide a link for online credit card payments. Of course, the more additional features are provided, the better.
It’s really difficult to make an educated selection, since the whole field is complex and intransparent. Like you said: One ends up testing a good looking solution with quite some effort before after a while an unforeseen roadblock appears.
One of my criteria is to use free open source software to avoid a lock in and have all options for creating integrations or additions in the future. This leads to some softer/non-technical selection criteria to avoid projects, which seem to be open source today, but are turned into commercial offerings with locked in customers tomorrow (thinking of the transition from OpenERP to the current state of Odoo, or what happened to OpenBravo and others).
Even for true open source solutions of today, this risk seems higher, if they are taken care of by a commercial company instead of a non-profit foundation. Like Frappe, the company behind ERPNext, is giving clear commitments to open source, on the other hand they just got a large investment and sometimes, this can lead to strategic changes (not implying that this is currently the case with Frappe, just highlighting a potential risk).
Finally, to have something relatively future proof, I was looking for solutions with a thriving community. Thriving in a niche field of ERP of course means something different than with other projects, but if the last forum post of a project is from six months ago, this is a red flag for me.
In the end I shortlisted solutions like Tryton, ERPNext, iDempiere, some others and Dolibarr.
To really have a certain degree of control over the solution, for me it’s important that the source code and architecture are accessible, i.e. it’s possible to understand the overall mechanics and be able to dig into details, if required (this is not quite an end user perspective, coming more from an IT background). In that regard, Dolibarr has the advantage of just using one programming language (PHP) for front end and back end and not requiring tons of external dependencies and complex frameworks. ERPNext as an example uses Python in the back end plus the vue.js framework and JavaScript for the front end, which is technically totally fine but adds another level of complexity which I prefer to avoid in this case.
Long story short, I got the impression that Dolibarr has a great potential for a long-term ERP solution, where a company wants to keep control over its core systems and data.
Regarding your help desk requirements, I am afraid I cannot provide any additional recommendations, since this is currently not a core issue I am dealing with. However, if the interface for external users does not meet your needs (or is problematic in terms of disclosing information), keep in mind that there is a good REST interface in place, which allows to create some own web pages with input forms for tickets or ticket lists and integrate them relatively easily with Dolibarr in the backend.