semantic/weekly/2016-09-09.md

# September 9th, 2016

- Hack week was last week and we skipped the weekly.
- We moved the weekly to the end of the week to try to cut down on the “wait, what _did_ happen last week?” thing.
- This ended up being @tclem’s first weekly 👋


#### What went well?

@joshvera:

Hack week:

- Made a lot of progress on the TypeScript parser.
- May be able to use that as the basis for a more rigorous JavaScript parser as well.

This week:

- Understand RWS & some other diffing algos a lot better than before.


@rewinfrey:

- Pairing w/ @tclem.
- Updating test cases is much more efficient.
- Modelling effects in Free.


@tclem:

- Little fixes.
- Static linking of ICU in dev.
- statsd client.


@robrix:

- semantic-diffd
- Docker
- kubes
- Markdown


#### What were the challenges?

@joshvera:

- Integrating a pass before RWS. Using constant-time (per-subtree)SES before RWS to match up equal things. There are some ordering problems with the result.
- Ambiguities in the TypeScript grammar. Possibly due to JS actually being context-sensitive. @maxbrunsfeld advises parsing a superset of the language… but which superset?


@rewinfrey:

- Trying to get an effect system in Free. Got it, but tricky.
- Also picked up an issue with template strings which defies debugging.


@tclem:

- > My head hurts by the end of the day.
- Being challenged by some of the new concepts, the vernacular &c.
- Hard to know when to jump down the rabbit hole and learn a thing or when to gloss over it.
- Have a queue of things to read.


@robrix:

- Converting between line/column ranges and character ranges.


#### What did you learn?

@joshvera:

- > I became one with the TypeScript grammar.
- > Maybe we should find a way to have the grammars write themselves.
- Working on this `effects` package (formerly called `freer`) & its `Eff` type. It uses an open union data structure, which `Data.Record` is sort of an approximation of. `Eff` has this detail where it’s list of function types are “type-aligned.” Every item in the list is a function type where they all chain together, a -> b, b -> c, etc. Adding, removing, & replacing effects is constant-time.

@rewinfrey:

- Free, and how to model effects in Free.
- Straight-up category theory stuff. Compositions of natural transformations &c.

@tclem:

- Post about functors, applicatives, and monads shown in pictures.
- How this manages to be a bridge between pure & stateful functions.
- Helped explain the optparse-applicative syntax.

@robrix:

- CMark exists, which is cool.
- Parsing/term ingestion had been producing errors for weirdly-formatted for loops &c. Not only were we hitting this on a regular basis, it was also causing confusing/poor change summaries.


#### Meta

“What were the challenges?” can be a bit redundant with “what did you learn?” Should we focus this on challenges that we need help with? Gonna give that a try.

Next week: @joshvera, @rewinfrey, & @robrix are off to ICFP. @tclem may or may not hold weekly solo at his discretion.