On Mon, May 25, 2026 at 02:14:14PM +0200, Jonas Smedegaard wrote: > [...] > > Fonts may have a preferred source form that is not textual but binary. > For another example of that, see etoys which is in non-free despite > being DFSG-free: It is kept out of main due to concerns that its source > format might be too alien for the Debian security team to reasonably do > audits on the codebase.
That seems an understandable concern for etoys, but not particularly relevant here - the SVG format is textual and easy to audit. > But from a legal standpoint, code either have freely licensed *source* > or they don't. And that applies to fonts in the sense that either the > font *is* its own source or it isn't. > > For font-awesome we cannot reasonably assume that the binary TrueType > code is the source for the code project. We have been told that none of the distributed versions of the font are the source for the fonts. But the distributed SVGs are licensed under CC-BY-4.0. And they could be used as source for a font. > Either what has been given to us includes the sources or it doesn't. > If it doesn't, then we should stop distributing it, because it is not > freely licensed. Or with the post 1998 terminology: It is not Open > Source if its source is missing. > > If someone then forks font-awesome and states that from now on, $FOO is > the preferred form for modification within the forked project, then the > question remains: how can that new project be freely licensed (a.k.a. > Open Source) if is is derived from a project where source was missing? It is a question, indeed, but as I'm not a lawyer, I can't answer that question. But this seems no different to the question of whether we can distribute an image that is licensed under an open source license but that was created using proprietary software. Perhaps the original source is in some proprietary format, but the exported (and subsequently distributed) image is in JPEG or similar format. Debian is full of images whose creation is not questioned. Best wishes, Julian
