On 2024-11-05 13:52, Jon Turney via Cygwin-apps wrote:
On 14/08/2024 08:17, Brian Inglis via Cygwin-apps wrote:
new source packages on pythonhosted appear to change dashes in package
names to underscores in archive names and directories, for example:
python-license-expression ->
license-expression ->
https://files.pythonhosted.org/packages/source/l/license_expression/license_expression-30.3.1.tar.gz
-> license_expression-30.3.1/
although dashes in pythonhosted package directory names
still seem to work as an alternative:
https://files.pythonhosted.org/packages/source/l/license{-,_}expression/license_expression-30.3.1.tar.gz
but dashes work and underscores do not in older package version
archives, although underscores in pythonhosted package directory names
still seem to work as an alternative:
https://files.pythonhosted.org/packages/source/l/license{-,_}expression/license-expression-30.3.0.tar.gz
Thanks for this.
Sorry about the delay in looking at this, but I have been very short of time
recently.
So, first off, I am a bit unhappy about guessing how this supposed to work. Have
PyPI not made any announcement of this change, explaining exactly what it is?
PEP 625 2020
https://peps.python.org/pep-0625/
normalization:
https://packaging.python.org/en/latest/specifications/binary-distribution-format/
TL;DR: distribution names cannot contain '-' as that is used to delimit version:
"In distribution names, any run of -_. characters (HYPHEN-MINUS, LOW LINE and
FULL STOP) should be replaced with _ (LOW LINE), and uppercase characters should
be replaced with corresponding lowercase ones.
This is equivalent to *regular* name normalization followed by replacing - with
_.
Tools consuming wheels must be prepared to accept . (FULL STOP) and uppercase
letters, however, as these were allowed by an earlier version of this
specification."
as opposed to project ("regular" above) name normalization rules:
https://packaging.python.org/en/latest/specifications/core-metadata/#core-metadata
https://packaging.python.org/en/latest/specifications/name-normalization/#name-normalization
So unnormalized names may work, or lowercased, or s/[-._]\+/-/, or s/[-._]\+/_/!
Secondly, I'm not crazy about the "try all the alternatives" approach. Perhaps
it would be better to form the URL using the current (new) rules, with an escape
(e.g. define PYTHON_ORG_OLD_STYLE_URL to form it using the old rules)?
https://warehouse.pypa.io/api-reference/legacy.html#upload-api
"All fields need to be renamed to lowercase and hyphens need to replaced by
underscores.
So instead of “Description-Content-Type” the field must be named
“description_content_type”.
Note that adding a field “Description-Content-Type” will not raise an error but
will be silently ignored."
https://github.com/pypi/warehouse/milestone/12
"Post Legacy Shutdown priorities
Important issues that have gotten unblocked now that legacy PyPI is dead (RIP).
See https://lwn.net/Articles/751458/ for more info."
https://lwn.net/Articles/751458/
Perhaps we should consider the Gentoo eclass changes:
"It should be noted that legacy URLs are no longer supported and may stop
working at an arbitrary time."
like recently:
https://projects.gentoo.org/python/guide/pypi.html
--
Take care. Thanks, Brian Inglis Calgary, Alberta, Canada
La perfection est atteinte Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer but when there is no more to cut
-- Antoine de Saint-Exupéry
