using ob-ipython as alternative of jupyter

ob-ipython introduce

ob-ipython repo

ob-ipython是基于org-babel,提供了一个在org-mode下做类似jupyter-notebook体验的开发模式. 用来快速做代码尝试或一个pandas数据分析很长合适.

为什么不使用原声的jupyter呢?

1. 基于browser的jupyter使用体验并不好.

按键配置和补全,snippets等等都是致命伤，这让jupyter notebook用起来更像个toy-project.

1. org-mode天生强大的功能配合ob-babel, ob-ipython. org-mode强大的文件编辑功能自不必说, 这种方式也比原生的代码里套markdown的方式更舒服. 配合上可以使用emacs所有的增强功能.

2. literal-programming的正确使用方式.

支持的特性

1. 内嵌pandas, matplotlib图片和表格数据
2. org-mode variable
3. 导出功能export
4. multi-language??
5. jupyter-kernel support??
6. working with a remote session??
7. 在代码中打开REPL to play with.
8. org literal-programming with :tang supported.

install in spacemacs

first add ob-ipython to spacemacs dotspacemacs-additional-packages

install jupyter

pip install jupyter

config ob-ipython in spacemacs `user-config`

(with-eval-after-load 'org
  (add-to-list 'org-babel-load-languages '(ipython . t))
  (add-to-list 'org-babel-load-languages '(rust . t))
  ;; 不再询问是否允许执行代码块
  (setq org-confirm-babel-evaluate nil)
  ;; displace/update images in the buffer after I evaluate
  (add-hook 'org-babel-after-execute-hook 'org-display-inline-images 'append))

add snipet for conversation

# -*- mode: snippet -*-
# name: ipython block
# key: srcip
# --
#+BEGIN_SRC ipython :session ${1::ipyfile ${2:$$(let ((temporary-file-directory "./")) (make-temp-file (concat (file-name-sans-extension (file-name-nondirectory buffer-file-name)) "-py") nil ".png"))} }:exports ${3:both} :results raw drawer
$0
#+END_SRC

use example and tips

使用maptlotlib inline来保证图片输出

%matplotlib inline
import matplotlib.pylot as plt
import numpy

plt.hist(np.random.randn(20000), bins=200)

使用ipyfile来控制生成的图片