Possible LTS releases tied to editions?

In general, yes. However, universe is also the place where Firefox-required toolchain backports go. For comparison, the gcc-mozilla package lives there, too.

That was meant as a positive statement about the less entitled attitude of RHEL users relative to the more entitled attitude of (some) Debian stable users. (Note users rather than maintaners in that sentence.) The logic goes that since Debian is a community project, the larger community should help enable Debian users to build new upstream software with the old toolchain in Debian stable. As noted, the relevance to this discussion is two-fold: 1) if an LTS isn't behind a commercial subscription, there are more people who use LTS without a very serious need to do so leading to LTS appearing as a bigger addressable audience creating more pressure to hold new software back in order to be compatible with the LTS and 2) when the LTS is perceived as a community effort, people feel it's more OK to ask the larger community to bear the externalities of the LTS.

My conclusion from this is that Rust should treat the C++ situation as something to avoid. It's super-sad that in 2018, when the latest GCC, clang and MSVC are a download away, stuff like Abseil enabling users of C++ stick to C++11 while backporting some newer features is perceived as a valuable thing.

That is, I think Rust should identify what it is that makes people shy to take a free-as-in-beer compiler upgrade up from C++11 and then avoid making the same mistake in the Rust context so that we can get users of Rust to keep upgrading rustc instead of the library writers diverting effort into enabling them not to.

6 Likes