On 30 December 2014 at 05:09, Chris Jerdonek <chris.jerdonek@gmail.com> wrote:

On Mon, Dec 29, 2014 at 12:01 AM, Nick Coghlan <ncoghlan@gmail.com> wrote:
> * for those cases (like date-based versions) where excluding releases with
> an additional numeric suffix is the right thing to do, an explicit prefix
> exclusion will still be possible, and will have the advantage of failing
> noisily on older versions of pip and easy_install, rather than silently
> installing an unexpected version (which is what will happen for anyone that
> starts relying on the current explicit exclusion in PEP 440). It's also
> possible to replace the exclusive bound with an inclusive bound that has the
> last release segment component incremented.

Just to confirm, in each of the below is it true that the two
comparisons are exactly equivalent (taking into account pre-release
and post-release rules, etc)?

1) >V.N.* and >=V.N+1
2) <=V.N.* and <V.N+1

If you omit the trailing ".*" (which isn't part of PEP 440 in the context of ordered comparisons), then the following are equivalent in the currently published version of the PEP:

1) "> V.N" and ">= V.N+1"
2) "<= V.N" and "< V.N+1"

The current behaviour of those exclusive comparisons is the part that's both incompatible with the previous behaviour of setuptools/pkg_resources, and somewhat confusing (largely because the interaction between the zero padding and the negative prefix matching means that the trailing zero in the release segment effectively isn't implied when comparing against maintenance releases).

Donald's proposed adjustment at https://github.com/pypa/interoperability-peps/pull/3 (which I now agree with) changes them to be different: unlike ">= V.N+1", "> V.N" will once again allow "V.N.x" maintenance releases, as it did in setuptools/pkg_resources releases prior to 8.0.

The exclusive ordered comparison behaviour in the originally accepted version of the PEP can still be requested explicitly as "< V, != V.*" and "> V, != V.*" (where the latter has the virtue of failing noisily on older versions of pkg_resources, rather than silently referring to a different minimum version).

And these two should be even easier:

3) <V.* and <V
4) >=V.* and >=V

With the way the PEP is written, it's not completely obvious to me
that these are true.

In a hypothetical future version of the PEP that allowed the ".*" wildcard suffix for ordered comparisons, it would likely have the following semantics:

"< V.*" ==> "< V, != V.*"

"> V.*" ==> "> V, != V.*"

"<= V.*" ==> "<= V"

">= V.*" ==> ">= V"

However, it's not clear that such an addition would be worth it, given the updated definition of exclusive comparisons, and the availability of the more explicit spelling.

Cheers,
Nick.

Nick Coghlan | ncoghlan@gmail.com | Brisbane, Australia