Revise the exclusive ordered comparison rules to no longer imply !=V.*#3
Conversation
|
Some minor editorial comments, and an actual error picked up by Marcus, but overall this looks good to me. I unfortunately renamed the PEP 440 file based on the discussion in issue #2, so that will also need changing in the PR before it will apply cleanly again. |
b7cabbd to
797df18
Compare
|
This is updated now. |
There was a problem hiding this comment.
Still another s/should/will/ here.
|
Generally looks good to me - just one change that was missed from the previous review, and a realisation that the change log could likely use an introductory paragraph. |
797df18 to
def6195
Compare
|
Updated again. |
|
Merging this one - I'll also add the 'c'->'rc' normalisation change to the changelog, and then give folks on distutils-sig one last chance to comment before we publish the change on python.org |
Revise the exclusive ordered comparison rules to no longer imply !=V.*
|
At what point can I merge the packaging PR and release it? |
Instead of having these operators imply a
!=V.*this will instead have them more-or-less simply check if the prospective version is greater than or less than the specifier version. They will not however match a post or pre-release of the version mentioned in the specifier unless the specifier itself contains a post or a pre-release. In addition, we will ensure that>Vdoes not matchV+local.versioneven though it is technically greater than V.In particular that means that something like
>1.7will match1.7.1but will not match1.7.post1while>1.7.post0would. Likewise<3.0would not match3.0.dev0while<3.0rc1would.This is implemented in pypa/packaging#25