What's the evidence to the contrary? When else has the FSF made a request that affects *development* as opposed to a release (other than licensing and issues affecting the principle of free software, which everybody agrees they have a right to)?
i386-unknown-lignux ; this broke unified tree builds since gcc wouldn't agree with binutils and newlib on the configuration. i386-pc-linux-gnu made it worse from a functionality standpoint because suddenly we went from host/target triplets to host/target quadruplets. i386-pc-linux_gnu would have fit much better in the old scheme. But the main issue back then was really the lack of coordination with other projects and the lack of a staged transition plan.