由于包冲突，无法使用conda安装basemap 1.2.1，如何解决？

Question

我在一个非根环境中的Conda（win1064，python3.7.3）上安装了Basemap，但最终出现了there is no epsg in the proj folder的问题。根据github的建议，我发现我的版本是1.2.0，并试图安装1.2.1，但没有成功。在

编辑：很明显，这是与proj不兼容的问题，在尝试此操作时可以看到：

conda create -n test python proj basemap=1.2.1 -c defaults -c conda-forge

首先，我将通道conda forge设置为最高优先级，我的环境的一半因此而更新，但是Basemap没有。在

然后我试图强制安装1.2.1，这会导致详细报告哪些软件包相互冲突：

Package llvm-meta conflicts for:
arrow-cpp -> numpy[version='>=1.16,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
parquet-cpp -> arrow-cpp=0.5 -> numpy[version='>=1.16,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
fastparquet -> numpy[version='>=1.11'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
statsmodels -> numpy=1.11 -> libblas[version='>=3.8.0,<4.0a0'] -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
pandas -> numpy[version='>=1.11.*'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
wordcloud -> numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
mkl-service -> numpy[version='>=1.11.3,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
bokeh -> numpy[version='>=1.7.1'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
blas -> openblas -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
nltk -> numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
liblapack -> libblas==3.8.0=14_openblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
matplotlib-base -> numpy[version='>=1.14.6,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
scipy -> blas==1.0=mkl -> libblas==3.8.0=10_mkl -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
libblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
scikit-learn -> libcblas[version='>=3.8.0,<4.0a0'] -> libblas==3.8.0=14_openblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
basemap=1.2.1 -> numpy[version='>=1.14.6,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
seaborn -> numpy -> libblas[version='>=3.8.0,<4.0a0'] -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
pyarrow -> numpy=1.11 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
libcblas -> libblas==3.8.0=14_openblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
matplotlib -> numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
patsy -> numpy[version='>=1.4.0'] -> libblas[version='>=3.8.0,<4.0a0'] -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
numba -> numpy[version='>=1.11,<1.12.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']
numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> llvm-meta[version='5.0.0|5.0.0.*']

Package clangdev conflicts for:
libblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
basemap=1.2.1 -> numpy[version='>=1.14.6,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
scikit-learn -> libcblas[version='>=3.8.0,<4.0a0'] -> libblas==3.8.0=14_openblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
matplotlib-base -> numpy[version='>=1.14.6,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
seaborn -> numpy -> libblas[version='>=3.8.0,<4.0a0'] -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
blas -> openblas -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
libcblas -> libblas==3.8.0=14_openblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
bokeh -> numpy[version='>=1.7.1'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
wordcloud -> numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
matplotlib -> numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
fastparquet -> numpy[version='>=1.11'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
numba -> numpy[version='>=1.11,<1.12.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
nltk -> numpy -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
scipy -> blas==1.0=mkl -> libblas==3.8.0=10_mkl -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
mkl-service -> numpy[version='>=1.11.3,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
pyarrow -> numpy=1.11 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
parquet-cpp -> arrow-cpp=0.5 -> numpy[version='>=1.16,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
pandas -> numpy[version='>=1.11.*'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
liblapack -> libblas==3.8.0=14_openblas -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
arrow-cpp -> numpy[version='>=1.16,<2.0a0'] -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
statsmodels -> numpy=1.11 -> libblas[version='>=3.8.0,<4.0a0'] -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']
patsy -> numpy[version='>=1.4.0'] -> libblas[version='>=3.8.0,<4.0a0'] -> libopenblas==0.3.7=h29e5d5d_0 -> libflang[version='>=5.0.0'] -> openmp==5.0.0 -> clangdev[version='5.0.0|5.0.0.*']

我不知道下一步该怎么做，环境中甚至都没有安装llvm-meta和{}，为什么会与不存在的包发生冲突？在

在安装basemap1.2.0时，也没有安装这些包，所以除非在版本1.2.1中更改了依赖关系，否则这对我来说毫无意义。在

我也不知道该向哪个github报告，任何建议都会有帮助。在

以下是conda list的输出：

^{pr2}$

Answer 1

你可能想破坏你的环境，重新开始。而且，看起来您几乎已经克隆了基本环境，您确定需要所有这些包吗？在

要移除环境：

conda env remove -n <env name>

要从头开始创建，可以指定多个频道进行搜索。conda forge频道有最新版本，但并不总是稳定的。建议尽可能使用默认频道。尝试使用以下方法创建环境：

^{pr2}$ 编辑

似乎有人在编译proj时没有检查依赖关系。如果你跑了

conda search proj=6.2.0  info -c conda-forge

它显示了一个- proj4 ==999999999999的约束。在