Currently at 129 and 519 are remaining. Five-One-Nine. OMG. I already feel like I encountered so many new people and products.

Also, Houston we have a problem.

The way stumble-twitter works is I have a file called 'twitterlinks.p' which acts like a storage for all the links and their concerned attributes. It's a relic that was passed down from the original find-birds code that Karpathy had written. It uses the pickle module to serialize/deserialize the data and I continued to use it.

So everytime I update a link with tags, I write to that file. Works perfectly on my system but as it turns out, the tags I make online at stumble-twitter will not persist in the current way that my code is written.

Heroku has an ephemeral file system. The changes made to files are reset every day/couple of hours with the dyno's restarting. The reason is that 'writing to local files is a bad development practice and Heroku does not allow it to force it's customers to follow good dev practices'.

Apparently, using a database would solve this issue.

SQLAlchemy. MongoDB. The choices are many. I've also heard of Redis as a possible solution.

I briefly looked at a Redis tutorial. The same things that threw me off Django are throwing me off this now. I'm new to databases. It's a new thing to learn. From the looks of it, it's unavoidable. I'll have to get to it some time, if not now then in the future. It's the next rung of the ladder.

Let's see.

I'll ask Bento if he has any pointers to give. Interesting how a single 'yeah, go ahead with xyz' from a mentor can make you so much more confident and push you over the edge of the initial learning inertia.