Siemens has closed access to OpenRadioss, the open-source finite element solver Altair Engineering released under the GNU Affero General Public Licence in 2022, according to a report from Phoronix. The project — an open-source edition of an industry-leading Altair solver — has been effectively withdrawn since Siemens acquired the company last year, with the report stating that access has been shut down.

For anyone building simulation, infrastructure and advanced-manufacturing workflows on vendor-sponsored open source, the episode is worth more than a headline about one discontinued solver. It is a clean, public case study in a governance failure mode that rarely appears on licence compliance checklists.

What actually happened

Altair open-sourced its Radioss finite element solver in 2022 under the GNU AGPL. OpenRadioss gave analysts and engineers a genuine option for large-scale explicit dynamics simulation without a proprietary licence.

That openness was conditional in practice, if not on paper. OpenRadioss was not administered by a neutral foundation, not governed by a multi-vendor consortium, and not underwritten by any contractual continuity commitment from Altair. It was a vendor-backed project whose future depended on a single corporate steward's incentives.

Siemens acquired Altair last year, and those incentives changed. Phoronix reports that access to the project has since been shut down.

Forkability in theory is not fork viability in practice

The common counter-argument to any withdrawal story is that the licence protects users: AGPL rights, including the right to modify and redistribute, survive the closure of upstream. On that reading, nothing substantive is lost.

That argument is only half true. OpenRadioss is not a lightweight utility a small group can carry over a weekend. It is a sophisticated solver codebase sustained by proprietary solver expertise, deep regression infrastructure and a contributor pipeline concentrated inside one company. No viable community fork has been confirmed, and whether any downstream group will carry the last public snapshot remains an open question; Phoronix's account of the repositories' inaccessibility has not been independently verified here.

For teams that adopted OpenRadioss, the exposure is immediate: loss of upstream patches, no supported path for solver improvements, and an audit trail that now terminates at a corporate decision rather than a community vote.

Why it matters in Hong Kong and elsewhere

Hong Kong's simulation-heavy sectors — engineering consultancy, infrastructure, advanced manufacturing — lean on open-source tooling partly for licence terms and partly for strategic appeal. The same is true of teams in Singapore, Shenzhen, London and beyond. Open source is not vendor-neutral by default. A project can be perfectly licensed and still be governed by exactly one company's commercial interest, and when that company is acquired, the openness can be reclassified internally within months.

The OpenRadioss case suggests the licence file is the weakest place to look for assurance. What matters more is custody: who holds steering rights, who employs the maintainers, and what happens upstream when a merger closes.

Four governance questions worth asking

  1. Stewardship model. Is the project administered by a neutral foundation or multiple independent maintainers, or does it sit inside one vendor's engineering org?
  2. Acquisition surface. Which upstream components your team depends on sit within a single corporate owner, and which of those owners are plausible acquisition targets?
  3. Forkability in theory vs. viability in practice. Can you realistically rebuild the solver, data pipeline and support function, or does the licence merely guarantee a theoretical right you would never exercise?
  4. Continuity language. Do contracts cover source escrow, notice periods before service withdrawal and a maintenance floor — or is continuity assumed rather than documented?

What the withdrawal does not prove

None of this shows that vendors should not open-source, or that open source is unsafe. What it shows is that licence freedom is not vendor independence. OpenRadioss will be judged by what community stewardship emerges next, not by what was taken away. The questions that matter for organisations in Hong Kong or elsewhere are about custody and governance — and they are worth asking before an acquisition closes, not after.

Note: This article is based on Phoronix's report on the OpenRadioss shutdown. The status of the project's repositories has not been independently verified at time of publication.


據 Phoronix 報道,Siemens 已終止外界對 OpenRadioss 的存取權。此專案為 Altair Engineering 於 2022 年以 GNU Affero General Public Licence 開源釋出的有限元素求解器,亦是業界領先的 Altair 求解器之開源版本。自 Siemens 於去年收購該公司以來,專案實質上已經停止運作,報道指其存取權已被封閉。

對於任何在廠商資助的開源專案之上,建立模擬、基建及先進製造流程的團隊而言,這宗事件的意義遠超過一則「某個求解器被停掉」的標題。它是一個清楚而公開的案例研究,揭示一種極少出現在軟件授權合規清單上的治理失敗模式。

事件真相

Altair 於 2022 年以 GNU AGPL 將其 Radioss 有限元素求解器開源。OpenRadioss 為分析師及工程師提供了真正可行的選擇,讓他們無需專有授權即可進行大規模顯式動力學模擬。

然而這種開放性在實質上是有條件的,即使表面看來並非如此。OpenRadioss 並非由中立基金會管理,亦非由多廠商聯盟治理,Altair 更沒有以任何合約形式承諾專案的持續性。它是一個由廠商贊助的專案,其未來取決於單一企業管理者的利益考量。

Siemens 於去年收購 Altair,相關利益考量隨即改變。Phoronix 報道稱,此後專案的存取權已被封閉。

理論上的可分叉性與實質上的可行性

任何「上游終止」的消息,最常見的反駁是:授權條款保障了使用者——AGPL 條款賦予的權利,包括修改及再發布的權利,並不會因為上游關閉而消失。從這個角度看,其實沒有任何實質內容真正損失。

這個說法只對了一半。OpenRadioss 並非一個小團隊在一個週末就能接手的輕量級工具。它是一個高度複雜的求解器代碼庫,需要依靠專有的求解器專業知識、深入的 regression 基礎設施,以及一條集中於單一公司內部的貢獻者渠道才能維持。目前未能確認為可行的社群分叉,任何下游團隊能否接手最後一個公開快照版本,仍是未知數;至於 Phoronix 對相關 repositories 已無法存取的描述,本刊並無獨立核實。

對於已採用 OpenRadioss 的團隊而言,影響即時而直接:失去上游補丁更新、求解器改進再無受支援的路徑,而審計軌跡如今終結於一項公司決定,而非社群投票。

對香港及其他地區的影響

香港高度依賴模擬的行業——工程顧問、基建、先進製造——對開源工具的倚賴,部分源於授權條款,部分則是出於策略考量。新加坡、深圳、倫敦及更廣泛地區的團隊亦是如此。開源並不意味著預設不受廠商主導。一個專案可以完全符合授權條款,卻仍由單一公司的商業利益所治理;而當這家公司被收購,開放性可能在數月之內於內部被重新定性。

OpenRadioss 的案例顯示,授權檔案(licence file)是尋求保障時最靠不住的地方。更關鍵的是「管治權」(custody):誰掌握專案的領導權,誰聘用維護者(maintainers),以及當合併完成時,上游會發生什麼事。

四個值得提出的治理問題

  1. 管理架構(Stewardship model): 專案是否由中立基金會或多位獨立維護者管理,還是困於單一廠商的工程部門之內?
  2. 收購風險面(Acquisition surface): 團隊所依賴的上游元件中,哪些掌握在單一企業所有者手中?這些所有者又有哪些是可能的收購目標?
  3. 理論上的可分叉性與實質上的可行性: 你能否在現實中重建求解器、數據管線(data pipeline)及支援功能?抑或授權條款只保障了一項你永遠不會行使的理論權利?
  4. 持續性條款(Continuity language): 合約是否涵蓋源碼代管(source escrow)、服務終止前的通知期及維護下限?還是持續性僅屬假設,從未成文?

撤回專案未能證明的事

以上種種並非證明廠商不應將專案開源,亦非證明開源不安全。它所揭示的是:授權自由不等於廠商獨立。OpenRadioss 的最終評價,將取決於隨後出現的社群管治模式,而非取決於它被奪走了什麼。對於香港或其他地區的機構而言,關鍵問題在於管治權與治理——而這些問題值得在收購完成之前提出,而非事後才追問。

註:本文基於 Phoronix 就 OpenRadioss 終止事件的報道撰寫。截至刊出時,專案 repositories 的狀態未經獨立核實。

新聞來源 / Original News Source