You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
# We do *not* call `DocMeta.setdocmeta!` with value `:(using AbstractAlgebra)`
# because we want to force that each jldoctest has its individual
# `setup = :(using AbstractAlgebra)`.
# The reason is that the AbstractAlgebra doctests are run also in Nemo,
# where `using AbstractAlgebra` must be avoided, as a few symbols change
# their meaning, e.g. `QQ`.
As a result we are duplicating this line again and again in src which is rather annoying:
Yet at least some of the AA doctests are also run by the Nemo/Hecke/Oscar CI. They now have a different setup block and as such, at least some of them run into test failures.
A potential fix might be to do something like this in Nemo (and then similar in Hecke, Oscar):
From
docs/make.jl
:As a result we are duplicating this line again and again in
src
which is rather annoying:Ideally we would replace the comment in
docs/make.jl
by something like this:But if we do that, then (at least in the past) the CI tests for Nemo, Hecke, Oscar may fail. This is because there this is done instead:
Yet at least some of the AA doctests are also run by the Nemo/Hecke/Oscar CI. They now have a different
setup
block and as such, at least some of them run into test failures.A potential fix might be to do something like this in Nemo (and then similar in Hecke, Oscar):
But someone needs to experiment with this, once there is an AA releases with
doctestsetup()
available.(And then we probably should also add
Hecke.doctestsetup()
)The text was updated successfully, but these errors were encountered: