Rewrite `sysconfig.get_config_vars` data generation to receive clearly-defined inputs

Feature or enhancement

Proposal:

Right now, some of the data is calculated during import, then we use that to calculate more data when building the sysconfig.get_config_vars cache. It would be good to transform all of this into a clearly defined algorithm, where we the data is generated based on a set of inputs, making it easier to generate data when cross-compiling, and allowing us to generate data for a specific environment (eg. source builds, vs installed build). This would have helped mitigate issues such as GH-127429, GH-128055, and GH-126789.

Additionally, this also gives us the opportunity to split the data into two sets: the minimum needed to evaluate the install scheme paths, and the non-critical additional data we expose in sysconfig.get_config_vars. This would help with the bootstrapping issue in python -m sysconfig --generate-posix-vars.

https://github.com/python/cpython/blob/f802c8bf872ab882d3056675acc79c950fe5b93c/Lib/sysconfig/__main__.py#L206-L221

Has this already been discussed elsewhere?

No response given

Links to previous discussion of this feature:

No response

Commencez par Lib/sysconfig/main.py, en particulier les lignes liées 206–221 et le flux python -m sysconfig --generate-posix-vars. Suivez la manière dont les données de configuration sont calculées lors de l’importation et de la génération du cache. La tâche est terminée lorsque l’algorithme de génération des données possède des entrées clairement définies, prend en charge des environnements distincts tels que la compilation croisée et les builds installés, et sépare les données du schéma d’installation des variables supplémentaires exposées.