12:02
<naugtur>
When I need to create a mental model of something, it helps to recreate it in a medium I can look at all at once. So I did: https://gist.github.com/naugtur/daf2aaf415e03e6c3ae5030e06f7b20e This is an attempt to visualize the change in Layer 0 refactoring.
12:03
<naugtur>

Having done that, I compared with something I tried to model earlier.

I compared the Layer0 refactor with the old refactor writeup from Mark Miller (https://github.com/erights/tc39pr-refactor-ModuleRecord) and the most significant difference is:

We separate into a distinct ModuleInitialization object the bookkeeping needed to guide module instantiation, linking, initialization, etc. Thus, once the initialization process completes, this bookkeeping state is no longer present. This helps us reason about post-initialization state separately.

Which was introduced to make CyclicModuleRecord cleaner. I think it could also remove the need for bidirectional reference between Module and SourceTextModuleRecord if introduced to Layer 0 refactor.

18:35
<Kris Kowal>
yulia littledan Do we have a convener for Module Harmony tomorrow?
18:37
<littledan>
I'm happy to run the meeting and take notes. Is anything more needed as part of "convening" it?
18:40
<Kris Kowal>
Not to my knowledge. I’ve started proposing topics on the agenda https://docs.google.com/document/d/1CD5lIBZLl24XBWbQhokqBdt4Zl7wPAcFJKJrgePr9HU/edit#
18:57
<Kris Kowal>
Also, recorded a reprise of my talk at plenary for anyone who missed that. https://youtu.be/NluxPr6eFVk
23:57
<Kris Kowal>
Aside: I want to write the phrase “link and load” even though the words are clearly in the wrong order.