Summary

  • PrimeTek moved future major versions of PrimeNG to a commercial licence while keeping existing MIT releases and security fixes available. ISC, which used PrimeNG widgets in Stork, chose the MIT-licensed Optimus UI fork for its DNS and DHCP management tool.
  • ISC says it will budget an annual contribution to the fork's maintainer, OpenNG, while experimenting with weekly dependency updates. A free licence settled the right to use code; it never hired the people who must keep that code fit for the next release.

The invoice was hiding in the button

Sorting a lease table looks like the least political act in network administration. Click the column heading; find the address; get back to work. Then the supplier of the widget that draws the heading changes its business model. Suddenly the control panel has a procurement meeting attached to it.

That is the unusually revealing detail in ISC's Stork 2.6 release, published on 15 October. Stork is a web tool for managing DNS and DHCP operations. The release adds consolidated lease tracking across Kea servers, OpenID Connect authentication, monitoring of BIND 9 zone transfers and support for Kea's configuration backend. In the same announcement ISC says it replaced the user-interface widget library beneath the tool. PrimeNG supplied the ordinary pieces: sorting, pagination, buttons. Ordinary pieces are where an operator spends a great deal of the day.

PrimeTek's own announcement explains the commercial turn. Its older MIT-licensed releases stay MIT, and the company says security fixes continue for those versions. Future major versions, beginning with PrimeNG 22, PrimeReact 11 and PrimeVue 5, move under PrimeUI licences. At launch the Commercial Suite costs $599 per developer through the end of 2026, then $799 from 2027; a further year of updates is offered at $399 per developer. A free Community licence remains for eligible users, including qualifying small organizations and nonprofits, with stated size and funding conditions. None of this rewrites the licence on a version already shipped. It changes the path to the next major version and who may maintain that path on what terms.

ISC did not treat that as an abstract culture war about open source. Its release lists the options it considered: keep selected PrimeNG components itself, move to Angular Material, refactor the interface in HTMX, or adopt a fork. It chose Optimus UI. The OpenNG project describes an MIT-licensed continuation of PrimeNG v21, with more than 80 Angular components. Its migration tool swaps packages and imports, then reports cases that still need human work. Its support page promises community support, with no service-level agreement. Those are concrete boundaries, not a magical “same thing, still free” sticker.

The cost of changing a library had already shown itself in ISC's 2025 development report. An earlier PrimeNG upgrade in Stork touched 329 files, with roughly 9,300 lines added and 9,300 removed. That was a different upgrade, a year before the current fork decision. It is a useful measure of how far a front-end dependency can reach into the application. A licence line in a package manifest can require a substantial engineering project in the product that depends on it.

ISC now says it is budgeting an annual financial contribution to OpenNG even though Stork is not yet earning it revenue. That sentence is worth more than a dozen declarations of community spirit. It names a payer. The amount and payment terms have not been published in the cited records; the commitment is a budget intention, not a receipt. ISC also says the team is experimenting with weekly dependency updates as vulnerability reports rise. Security-only fixes on the old PrimeNG branch remain available, but a team building new Stork features needs a maintained route through framework changes, bugs and component behaviour.

A patch channel and a development future are different products.

The fork has a governance question of its own. OpenNG can publish code; ISC can adopt it; Stork operators can depend on it. Those acts do not give every operator a vote over OpenNG's release calendar, a guaranteed fix deadline or the ability to move the entire interface overnight. The maintainer bears triage and release work. ISC bears integration, testing and support to its users. DNS and DHCP administrators bear the cost if either layer stops matching the other. Calling everyone “the community” makes the table look crowded while leaving the invoice under the napkin.

There is a sensible reason PrimeTek moved to paid licensing. It says its add-on sales no longer funded the engineering, accessibility, documentation, framework compatibility and support expected of four large libraries. Its price and licence terms are a claim about how PrimeTek intends to fund that work. The fork is a competing answer to the same maintenance question. Neither answer becomes durable because a README uses the word sustainable. The proof is release discipline, vulnerability handling, migration cost and money arriving where the work is done.

For operators, this is the part of a software bill of materials that a dependency scanner cannot settle alone. The scanner identifies a package and version. It does not decide whether the maintainer has enough time, whether a fork can absorb upstream security changes, who pays for test coverage, or how long a critical control screen may lag a framework update. Those are purchasing and governance decisions even when the purchase order says zero.

A button is a small place to hide a veto

The Stork case exposes a familiar asymmetry. The organization shipping the network tool chooses its UI dependency, yet users inherit the operational consequences of that choice. A vendor can change terms for future releases; a fork can offer continuity; neither move automatically funds the developer who will reconcile the next broken component with a DHCP operator's working day. ISC has at least made its answer visible: move to an MIT fork, plan to contribute money, and update dependencies more often. The next test is whether that promise survives the dull work after the launch post.