Skip to content

Latest commit

 

History

History
43 lines (40 loc) · 3.36 KB

NOTES.md

File metadata and controls

43 lines (40 loc) · 3.36 KB

Scratchpad of notes and thoughts that occurred to me throughout development, or helpful things to remember

  • At present (before it's generated dynamically), the URL for the statically-hosted S3 file is http://pop-culture-graph.s3-website-us-east-1.amazonaws.com
  • To create the stack, call: aws cloudformation create-stack --stack-name test --template-body file://template.yaml --parameters ParameterKey=paramGithubOAuthToken,ParameterValue=[REDACTED] ParameterKey=paramGithubRepo,ParameterValue=popCultureGraph ParameterKey=paramGithubUser,ParameterValue=scubbo --capabilities CAPABILITY_NAMED_IAM
  • Open question here on how best to build SAM applications.
  • Not necessary in this case (since the website can just call a Route 53 DNS name), but I'd be interested to know the best solution for if the static website needed to reference a Cloudformation Stack Output (presumably - "it shouldn't, just have it call some backend service that references the Stack Output")
  • The documentation on how importing works in Python is lacking (or, I can't find it): ** help();help(import) gives The details of the first step, finding and loading modules are described in greater detail in the section on the import system. ** help();topics lists several topics that look potentially helpful: IMPORTING, MODULES, PACKAGES *** IMPORTING contained the same content (at a skim - specifically, it contained the same "The details of the first step..." section) as help(import) *** MODULES only contained information for interacting with (already-imported) modules *** PACKAGES containes the same content as help(import) and IMPORTING
  • I'm really lacking a neat way of referencing a common root (such that a test runs with the same relative working dir whether run from PyCharm or from pytest .). A variable like package-root would be useful (probably exists, I just haven't looked it up yet)

TODO

  • Specific unit-tests for DataFetcher, not just app-level
  • requests-mock tests for DataFetcher
  • Make the bucket-name in buildspec.yml dynamic {BUCKET-NAME}, rather than hard-coded: like in here
  • Deploy static website in a Deploy step, rather than during build
  • Framework to test locally (website calling local functions)
  • Notification when pipeline fails
  • Rewrite from line 313-onwards of main.js, which handles responses from clicking nodes in spread mode. Update to match the current data-format (and/or update the data-format to be more helpful - might be nice to make the responses statelessly-consumable, without requiring the client to track that!)
  • Do not duplicate existing linked-nodes (probably requires passing a list of "nodes to suppress" into the functions)
  • franchises_for_actor for Alison Brie (1555340) is returning empty ditto Zoe Saldana
  • Many incorrect character-name parsings (e.g. ) as a character name of Charlie Day) Also Will Arnett has two "Riviera" entries
  • Clicking a node N_1 that's already connected to a node N_2 will reset the chunkNum of N_2
  • When pruning, also remove all leaves
  • Clicking spread one one node, then on another one quickly afterwards (before completion) leads to the resulting three nodes being "linked" to nothing