Uploaded with version adjusted to 9.5-6ubuntu1 (just to avoid one of the lintian warnings about a missing version).
I had a look through the test failures and quite a lot of them are a case of "output when not expecting any", with the output being deprecation warnings largely concerning pkg_resources and numpy types (which are not terribly surprising). It appears this is the bulk of the "extra count" required (comparing to the current archive build).
Other errors are cases of reductions changing (because sympy changed?) for certain things. For instance:
Still others are errors that do look genuinely "wrong" to my eyes:
File "src/sage/env.py", line 16, in sage.env
Failed example:
out == repr((SAGE_ROOT, SAGE_LOCAL)) # long time
Expected:
True
Got:
False
But I checked several of these cases against the current build in the archive, and these are not "new" either, so I'm reasonably satisfied the bulk of the extra error count is from new deprecations producing unexpected output for certain tests.
Uploaded with version adjusted to 9.5-6ubuntu1 (just to avoid one of the lintian warnings about a missing version).
I had a look through the test failures and quite a lot of them are a case of "output when not expecting any", with the output being deprecation warnings largely concerning pkg_resources and numpy types (which are not terribly surprising). It appears this is the bulk of the "extra count" required (comparing to the current archive build).
Other errors are cases of reductions changing (because sympy changed?) for certain things. For instance:
File "src/sage/ calculus/ calculus. py", line 1811, in sage.calculus. calculus. inverse_ laplace laplace( 1/s^2*exp( -s), s, t, algorithm='sympy')
Failed example:
inverse_
Expected:
-(log(e^(-t)) + 1)*heaviside(t - 1)
Got:
(t - 1)*heaviside(t - 1)
Still others are errors that do look genuinely "wrong" to my eyes:
File "src/sage/env.py", line 16, in sage.env
Failed example:
out == repr((SAGE_ROOT, SAGE_LOCAL)) # long time
Expected:
True
Got:
False
But I checked several of these cases against the current build in the archive, and these are not "new" either, so I'm reasonably satisfied the bulk of the extra error count is from new deprecations producing unexpected output for certain tests.