[llvm-dev] What version comes after 3.9? (Was: [3.9 Release] Release plan and call for testers)

Mehdi Amini via llvm-dev llvm-dev at lists.llvm.org
Fri Jun 24 22:01:17 PDT 2016

> On Jun 24, 2016, at 7:41 PM, Hans Wennborg <hans at chromium.org> wrote:
> On Mon, Jun 13, 2016 at 4:54 PM, Hans Wennborg <hans at chromium.org> wrote:
>> Breaking this out into a separate thread since it's kind of a separate
>> issue, and to make sure people see it.
>> If you have opinions on this, please chime in. I'd like to collect as
>> many arguments here as possible to make a good decision. The main
>> contestants are 4.0 and 3.10, and I've seen folks being equally
>> surprised by both.
> Thanks everyone for chiming in.
> Please correct me if I misrepresent your opinion here, but I need to
> try and summarize this thread for my own sanity:
> The thread started out with lots of support for 3.10, the reasoning
> being roughly that we shouldn't bump the major version number unless
> we want to signify major change (Mehdi, Hal, Blaikie, Saleem,
> Chandler, Anton, Eric, Aaron, Sean, Vikram).

Since I’m mentioned here in support of 3.10, I’d like to correct by repeating what I wrote previously in this thread:

"To clarify my point: I don't have a particular opinion about bumping the major number for whatever other reason than breaking the compatibility, but I'd probably suggest that we rewrite the compatibility policy to say something like "The current LLVM version support loading any bitcode since version 3.0”.”

Otherwise either Rafael last proposal (starting at 4.0 and +1 for every release), or Dmitri (“ubuntu-like” date-based numbering) are perfectly fine for me.


> Richard suggested that since we do time-based rather than
> feature-based releases, the distinction between a release with or
> without major changes is arbitrary, and we should move to a scheme
> where we update the major version number on each release (4.0, 5.0,
> etc.) with minor releases in between (4.1, 5.1, ..).
> Chris advocated for "keep adding 0.1 to each major release" (in the
> decimal sense), i.e. 3.9, 4.0, 4.1, etc. I haven't seen anyone else
> suggest this. "I do not think it is reasonable at all to go to '3.10'
> after '3.9', because we will never get to '4.0'."
> Chris then expressed support for alternatively just incrementing the
> major version each time, as Richard suggested, but starting at 40.
> Rafael expressed support for the above, but starting at 4.0: "It is
> simply not worth the time to try to figure out what is 'major' in a
> project with so many different uses."
> Chandler said he didn't like Chris's "keep adding 0.1 to each major
> release" scheme: "we shouldn't just go from 3.9 to 4.0 because of some
> decimal correspondence", and said he was open to either going to 3.10
> with the current major/minor split, or if we don't want that, use
> Richard's suggestion.
> Michael pointed out that if we do change the numbering scheme,
> changing the binary compatibility guarantee to something time-based
> isn't equivalent to what we currently have.
> So, it seems we're at an impasse with several folks in favour of 3.10,
> Chris speaking out strongly against it, and Richard's option which has
> some traction and which no one's disagreed with so far, but which
> would be a bigger change.
> I'll have a think about this over the weekend.
> Cheers,
> Hans

More information about the llvm-dev mailing list