[llvm-dev] (no subject)

Renato Golin via llvm-dev llvm-dev at lists.llvm.org
Wed Jun 15 04:03:05 PDT 2016

On 15 June 2016 at 03:51, Adve, Vikram Sadanand via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
> However, major version numbers also have a communication value: signifying a major step forward in the system along some dimension.

That was David's point, and I agree.

> but regardless of what the change is, I do think there is some value in reserving major version increments (like 3.xyz to 4.xyz) for major changes.  3.9 is not a fraction -- if it was, we would not have 3.7.1 etc.

That, OTOH, is not really relevant. There's no reason 3.9 can't be a
fraction, but we're getting into bike shedding and I don't think we

I personally don't have any infrastructure or scripts that will be
affected by either 4.0 or 3.10, so I don't really mind either way. But
using 3.10 would be *much* better for other OSS users and OS
distributions, so that's a very valid argument.

To end what has become a bit pointless discussion, I think we should
take an executive decision. There are more people in favour of 3.10
than 4.0, which seems we're building a consensus.

So... unless someone (upstream or downstream) can present a really
relevant and important point to *not* use 3.10, we should just go with
it, and continue the trend until we have some really major change.


More information about the llvm-dev mailing list