Re: Switching to PEP440 compliant versioning for our development code
Peter Cock <[email protected]>
| Newsgroups | gmane.comp.python.bio.devel |
|---|---|
| Message-ID | <CAKVJ-_6jo3ZRbiT+83NYfbdg-nYa=PDdYB+PaE7FzRT4mgSp5w__7779.36863382838$1466695149$gmane$org@mail.gmail.com> |
I meant to send this to the dev mailing list, CC'd, but any comments from the wider Biopython community are also welcome. Peter On Thu, Jun 23, 2016 at 2:14 PM, Peter Cock <[email protected]> wrote: > Dear Biopythoneers, > > Currently we use a simple two level versioning scheme > of 1.N for Biopython (currently at 1.67), which is all fine for > the recent PEP standards which are important for when > we release code via PyPI, and determining dependencies > and ordering releases. > > However, as currently documented on our release process, > in between releases we append a plus, currently 1.67+, > and these versions are not PEP440 compliant. > > http://biopython.org/wiki/Building_a_release > > The closest equivalent in the standard to 1.67+ (which we > are using to mean after 1.67 but before 1.68) would be > a post release, e.g. 1.67.post > > https://www.python.org/dev/peps/pep-0440/#post-releases > https://www.python.org/dev/peps/pep-0345/#version-specifiers > > However, having read PEP440, I think it would be clearer > to use 1.78.dev (a development pre-release of what will > be 1.68) instead? > > This still allows for alpha, beta and release candidates > should we need them, e.g. 1.78.dev, 1.78a, 1.78b, 1.78rc, > 1.78 etc. > > i.e. Rather than currently using 1.67, 1.67+, 1.68, 1.68+, ... > I am proposing we use 1.67, 1.68.dev, 1.68, 1.69.dev, ... > > Note that pushing development releases to PyPI is frowned > on, and I am not advocating we do that. This should only > be of interest for people who have installed from our git > development repository. > > Peter