On Tue, Jun 23, 2026 at 12:16:56PM +0100, Stuart Henderson wrote:
> https://github.com/numpy/numpy/releases/tag/v2.5.0
> 
> excerpt from the top:
> 
> "Numpy 2.5.0 is a transitional release. It drops support for Python
> 3.11, marking the end of distutils, and expires a large number of
> deprecations made in the 2.0.x release. It also improves free threading
> and brings sorting into compliance with the array-api standard with the
> addition of descending sorts. There is also a fair amount of preparation
> for Python 3.15, which will be supported starting with the first rc."
> 
> Given the deprecations, if we're wanting to move to 2.5+ in this release
> cycle, we want this sooner rather than later so that we have more chance
> to fix problems.
> 
> Test results so far:
> 
> amd64:
> old: = 40 failed, 47208 passed, 1790 skipped, 2817 deselected, 32 xfailed, 2 
> xpassed, 14 warnings in 1386.01s (0:23:06) =
> new: = 40 failed, 47051 passed, 1749 skipped, 3634 deselected, 32 xfailed, 2 
> xpassed, 14 warnings in 259.71s (0:04:19) =
> 
> aarch64:
> old: = 99 failed, 46433 passed, 394 skipped, 2817 deselected, 32 xfailed, 2 
> xpassed, 76 warnings in 638.22s (0:10:38) =
> new: = 99 failed, 46276 passed, 353 skipped, 3634 deselected, 32 xfailed, 2 
> xpassed, 76 warnings in 258.86s (0:04:18) =
> 
> amd64 tests were run on a busy machine so I wouldn't pay too much
> attention to timings there, however aarch64 was otherwise idle
> and that's quite a serious improvement...

Ran through an amd64 bulk (excluding a handful of monsters that I never
build because it's a waste of time and electricity)

ok tb

On sparc64 I get (I doubt it's some rather slow tests among the ~150
fewer passing ones that result in the massive timing differences we see
on the three architectures):

==> numpy-2.4.6-test.log <==
= 87 failed, 45706 passed, 1133 skipped, 2817 deselected, 32 xfailed, 2 
xpassed, 24 warnings in 1858.60s (0:30:58) =

==> numpy-2.5.0-test.log <==
= 87 failed, 45549 passed, 1092 skipped, 3634 deselected, 32 xfailed, 2 
xpassed, 24 warnings in 1144.93s (0:19:04) =

  • numpy 2.5.0 Stuart Henderson
    • Re: numpy 2.5.0 Theo Buehler

Reply via email to