Posted Mar 9, 2011 10:49 UTC (Wed) by paulj (subscriber, #341)
Parent article: Red Hat and the GPL
I'd like to restate my comments on earlier points. These points are speculative/exploratory, however I do think it's important for the community to be careful about ceding to a narrow interpretation of the GPL mostly out of respect for a friend. The friend may be trust-worthy and generally do the right thing, but others may not be.
Firstly, re "preferred form". The article makes the claim that RedHats' preferred form is not a factor for the GPL, but I can not find anything in the GPLv2 text to back this up. Surely the distributor's preferences for the form to work on *would* be a major factor in deciding whether the released source meets the "preferred form for modifications" criteria of the GPL? This part of the article seems very weak and poorly substantiated (if not outright wrong).
It used to be that RedHats' source for the kernel was a src.rpm with a base kernel and a series of patches, which was the source input to the build process for the kernel binary rpm. To me, it's fairly clear the GPL's definition of "source" would require that original src.rpm to be released, rather than any intermediate, auto-generated form with all the patches collapsed. However, it could be their process is now different and there no longer is such a "split patches" src.rpm. It could be their processes now are to build directly from the head of a git tree, in which case RedHat may be fine as this significantly decouples the history from the build process, and history is not source - though their reputation in the community may still suffer.