|
2772 | 2772 | "name": "stdout",
|
2773 | 2773 | "output_type": "stream",
|
2774 | 2774 | "text": [
|
2775 |
| - "Wall time: 6.8 s\n", |
2776 |
| - "Wall time: 5.26 s\n", |
2777 |
| - "Wall time: 345 ms\n", |
2778 |
| - "Wall time: 674 ms\n", |
2779 |
| - "Wall time: 672 ms\n" |
| 2775 | + "pickle.gz: Wall time: 6.77 s\n", |
| 2776 | + "csv.gz: Wall time: 5.45 s\n", |
| 2777 | + "hdf (full): Wall time: 664 ms\n", |
| 2778 | + "hdf (appd): Wall time: 840 ms\n" |
2780 | 2779 | ]
|
2781 | 2780 | }
|
2782 | 2781 | ],
|
2783 | 2782 | "source": [
|
| 2783 | + "print('pickle.gz: ', end=' ', flush=True)\n", |
2784 | 2784 | "%time graph_info_df.to_pickle('datasets/git-commit_graph.info.pickle.gz')\n",
|
| 2785 | + "print('csv.gz: ', end=' ', flush=True)\n", |
2785 | 2786 | "%time graph_info_df.to_csv('datasets/git-commit_graph.info.csv.gz')\n",
|
2786 |
| - "try:\n", |
2787 |
| - " #%time graph_info_df.to_feather('datasets/git-commit_graph.df_edgelist.feather')\n", |
2788 |
| - " %time graph_info_df.to_parquet('datasets/git-commit_graph.info.parquet')\n", |
2789 |
| - "except ImportError:\n", |
2790 |
| - " print(\"Missing optional dependency 'pyarrow' required\")\n", |
| 2787 | + "print('hdf (full):', end=' ', flush=True)\n", |
2791 | 2788 | "%time graph_info_df.to_hdf('datasets/git-commit_graph.info.hdf5', 'df', mode='w', complevel=6)\n",
|
| 2789 | + "print('hdf (appd):', end=' ', flush=True)\n", |
2792 | 2790 | "%time graph_info_df.to_hdf('datasets/git-commit_graph.hdf5', 'df_info', mode='a', complevel=6)"
|
2793 | 2791 | ]
|
2794 | 2792 | },
|
| 2793 | + { |
| 2794 | + "cell_type": "code", |
| 2795 | + "execution_count": null, |
| 2796 | + "metadata": {}, |
| 2797 | + "outputs": [], |
| 2798 | + "source": [ |
| 2799 | + "# feather does not support non-numerical index, IIRC\n", |
| 2800 | + "#%time graph_info_df.to_feather('datasets/git-commit_graph.info.feather')" |
| 2801 | + ] |
| 2802 | + }, |
| 2803 | + { |
| 2804 | + "cell_type": "code", |
| 2805 | + "execution_count": null, |
| 2806 | + "metadata": {}, |
| 2807 | + "outputs": [ |
| 2808 | + { |
| 2809 | + "name": "stdout", |
| 2810 | + "output_type": "stream", |
| 2811 | + "text": [ |
| 2812 | + "Wall time: 423 ms\n" |
| 2813 | + ] |
| 2814 | + } |
| 2815 | + ], |
| 2816 | + "source": [ |
| 2817 | + "%%time\n", |
| 2818 | + "try:\n", |
| 2819 | + " graph_info_df.to_parquet('datasets/git-commit_graph.info.parquet')\n", |
| 2820 | + "except ImportError:\n", |
| 2821 | + " print(\"Missing optional dependency 'pyarrow' required\")" |
| 2822 | + ] |
| 2823 | + }, |
2795 | 2824 | {
|
2796 | 2825 | "cell_type": "code",
|
2797 | 2826 | "execution_count": null,
|
|
0 commit comments