Cliff of Confusion

I have had some encouraging chats on the slack channel (that Zac of turned me on to), basically describing the Cliff of Confusion I am about to face with the impending end of Freecodecamp. It is a very real and troubling thing but I will power through it.


yesterday I just about managed to recreate the meteor-react tutorial which wasn’t really progress but it is a start. time to adapt meteor to booktrading needs.

as a reminder, deploying meteor to heroku requires these steps: I forgot to add the buildpack before instantiating the heroku instance, and had to figure out how to bump the package version to force the rebuild in heroku and npm version patch doesn’t seem to work for me. I dont know why but just manually editing minor things in the files worked to force the rebuild.

10.30 here. I have deployed the meteor app to heroku and that took embarrassingly long because I had forgotten so much. Now I am ready to start mutating it.

User Stories from

  • User Story: I can view all books posted by every user.
  • User Story: I can add a new book.
  • User Story: I can update my settings to store my full name, city, and state.
  • User Story: I can propose a trade and wait for the other user to accept the trade.

First off, this demo app sucks (what happens when a trade is accepted? god knows) so I am not going to bother replicating this UI or UX. But it will do me good to think about the data model and the clientside needs. Here are the implementation levels I am thinking:

  • MVP: User signup with settings to store name, city, state, and user books.
  • Stage 2: View all books.
  • Stage 3: Propose trade.
  • Stage 4: Accept trade.

1.40 here. meteor turned out to be far deeper than i thought. I started on this good video series: before realising he was teaching Blaze, and then did a bunhc of reading into Blaze vs React on Meteor (which is a rabbit hole of this, that, and “react is the worst”