Whether to clear cache

[luzai]

In item_recommendation, in the training process of discriminator, it seems need to clear the cache of old file, since you use linechache in ut.get_batch_data.

for d_epoch in range(50):
    if d_epoch % 5 == 0:
        generate_for_d(sess, generator, DIS_TRAIN_FILE)
        # whether to clear cache for old file? It seems need to add below two line code.
        import linecache
        linecache.clearcache()
        train_size = ut.file_len(DIS_TRAIN_FILE)
    index = 1
    while True: 
        # ut.get_batch_data to generate a minibatch
        # train discriminator for this minbatch

[LantaoYu]

Based on linecache documents, I think we don't need to explicitly clear cache to ensure correctness and clearcache() is used only when you no longer need lines from files previously read using getline(). But if there is indeed any error, any pull request is welcome.

[AaronHeee]

Hi authors,

I test linecache using some simple code in both python 2/3, finding that: If we don't use clearcache(), dis-train.txt will only use the sampled negatives in the first iteration. It seems the updated G will not affect D. Please let me know if I misunderstand something. Thanks!

The code snippet runs as follows:

>>> with open("test.txt", "w") as f:
...     f.write("01234")
... 
>>> linecache.getline('test.txt', 1)
'01234\n'
>>> with open("test.txt", "w") as f:
...     f.write("56789")
... 
>>> linecache.getline('test.txt', 1)
'01234\n'
>>> linecache.clearcache()
>>> linecache.getline('test.txt', 1)
'56789\n'