@Yvan, you donāt see a possibility of relaxing the GPLv3 OSS license to another, still compatible with GPLv3 but not as draconian and limiting the developerās options, do you? As an example of my own situation, Iām experimenting with sound processing just for the fun of it, and am using Intel One libraries, which are de facto the golden standard in high-performance computing, including DSP. They are closed-source libraries, and are incompatible with any variety of GPL, with a possible (IANAL, though) exception of LGPLv2. Certainly not with GPLv3. A large part of the OSS community has been reluctant to adopt GPLv3 because of its clear imbalance on interoperability. The most common OSS license used to-day is Apache 2.0, ācompatibleā with GPLv3, although without a legal precedent the definition of ācompatibilityā varies with the legal scholar you ask; itās fair to call it a gray legal area. Some solicitors even hold the opinion that compiling a GPLv3 source with a non-OSS toolchain, such as that which comes with Microsoft Visual Studio, is impermissible because it constitutes illegal combining GPL-licensed source with a closed-source part of the toolchain, C/C++ runtime library to produce a runnable binary.
Apache 2.0 is commonly preferred over the MIT license because it provides a strong protection against patent trolls, like GPLv3 and unlike the MIT. Both Apache 2.0 and MIT licenses are permissive, in the sense that they allow the use of OSS code in commercial applications. Naturally, such a commercial use would require the adopter of the VST3 OSS code comply with Steinberg and Intel licenses.
GPLv3 prohibits release of the final product in any form, source or binary, if it is linked with Intelās oneIPP. This is a bummer. Of course, implementing the function provided by the Intelās library is not something that I would consider even in a nightmare.
In my opinion, shared to at least a noticeable degree within the modern OSS community, the FSF philosophy and its family of GPL licenses fell behind the times and is archaic for the modern OSS. Weāre no longer living in the Bell Labs times, where engineers were hired and then given a complete freedom to work on what they find interesting and promising.
While for-profit businesses are often open to the idea of OSS, none of them in their sane mind would release code under any of the GPL varieties, for its infectiousness and incompatibility with many other libraries, be they OSS or not. The unfortunate and unexpected non-repeal of the Section 174 in December 2023 aggravates the situation manifold: the wave of bankruptcies of small software business, practical inability of starting a software business in the US, and the coming significant reduction of software engineering workforce in large businesses, all of which are the inevitable consequences of this ill-conceived law coming into effect the last year, and the current dysfunction of the US Congress leaves little in the way of a possible resolution of this new grave problem.
It must be noted that GPL licenses are acceptable in narrow niches, such as stand-alone executable programs, but itās hard to see how they pragmatically sensible to apply outside of these. The decision of the Linux kernel team not to adopt GPLv3 is telling a lot. Mac OS is based on FreeBSD chiefly because of legal impossibility to adopt the Linux kernel for it because of its GPL licence.
The bottom line is the GPL licenses pragmatically outlived themselves as the default to-go options, save for narrow circumstances. Iāve avoided touching on other reasons outside of purely rational and pragmatic ones very explicitly; in particular, folkās philosophical position that the level of software freedom enshrined by the FSF in the GPL family of licenses trumps absolutely any possible arguments, and am kindly asking the discussion participants to stay in cold mind and within the pragmatic reasoning scope strictly along the line of reasoning that I exposed above. In my 40 years in the field Iāve witnessed countless ālicense wars,ā and the one waged in this discussion would be utterly non-constructive, to say the least. On the other hand, arguments exposing possible flaws within the scope of my rational line of reasoning are welcome indeed.
Most of us are musicians and software engineers creating OSS software in our own free time, not barristers. Absolutely the most of us are wanting legal education even to determine whether releasing OSS VST3 software is permissible. GPL licenses are both notoriously voluminous and written in impassable legalese, which would require us to pay an arm and a leg for the solicitorās services, which doesnāt even add the value commensurate with the cost and the hassle, as GPLv3 is legally in a gray area, lacking, as I already mentioned, court decisions, and legal opinions on it vary. Additionally, the FSF is in fact known for prosecuting people for license non-compliance. I certainly wonāt risk grasping the wrong end of the stick to my financial ruin. To me, the choice of GPLv3 for OSS VST3 release is precluded, and Steinberg currently offers no alternative: I may sign the licensing agreement (as I once did for VST2 SDK around 2005, when this was the only option) and choose the non-GPL license option, but this also wonāt permit me release my work as OSS.