On 8 September 2015 at 02:22, Sven R. Kunze firstname.lastname@example.org wrote:
On 07.09.2015 02:26, Nick Coghlan wrote:
For the build farm idea, it's not just writing the code initially, it's operating the resulting infrastructure, and that's a much bigger ongoing commitment. Automatically building wheels for source uploads is definitely on the wish list, there are just a large number of other improvements needed before it's feasible.
Could you be more specific on these improvements, Nick?
- PyPI: migrating from the legacy Zope codebase to Warehouse - PyPI: end-to-end content signing (PEPs 458 & 480) - PyPI: automated analytics & dashboards - Tooling: integration with operating systems & other platforms - Python Software Foundation financial sustainability - Python Software Foundation project management capacity - Infrastructure improvements for the CPython workflow
Those aren't dependencies of automatic wheel-building per se, but rather are issues that are higher priorities for folks like Donald (in terms of actually getting things done), myself (in terms of collaborating more effectively with other open source ecosystems), and the PSF staff and Board (in terms of ensuring the python.org infrastructure is being appropriately maintained).
Running an automated build service is expensive, not primarily in setting it up, but in terms of the ongoing sustaining engineering costs (including security monitoring and response), so before we commit to doing it, we need to know how we're going to fund it. However, most of the PSF's focus at the moment is on getting the things we *already* do  on a more sustainable footing, so adding *new* services isn't currently a priority.