Skip to content

Commit a5b858a

Browse files
committed
add 2019-02-05 minutes; closes nodejs#16
1 parent d3f1f1c commit a5b858a

File tree

1 file changed

+108
-0
lines changed

1 file changed

+108
-0
lines changed
+108
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,108 @@
1+
# Node.js Foundation Tooling Group Meeting 2019-02-05
2+
3+
## Links
4+
5+
- **Recording**: [YouTube](https://youtu.be/fhFOrO-IEdc)
6+
- **GitHub Issue**: [#16](https://github.com/nodejs/tooling/issues/16)
7+
- **Minutes Google Doc**: [Google Doc](https://docs.google.com/document/d/1YQ9Z4Tr_sXUS4YbjtlIsUOJTKoRnKPaCEp29W9rX29w/edit)
8+
9+
## Present
10+
11+
- Tooling team: @nodejs/tooling
12+
- Ben Coe
13+
- Daniel Ehrenberg
14+
- Christopher Hiller
15+
- Ian Sutherland
16+
- Refael Ackermann
17+
- Michel Lopez
18+
19+
## Agenda
20+
21+
- Daniel Ehrenberg from TC39 wants feedback from the broader community
22+
- Daniel would give us some details on stuff we’re interested in
23+
- Ben Coe is interested in transpilers, source maps
24+
- Daniel:
25+
- no solid spec for source maps
26+
- there’s a summary available from TC39 meetings
27+
- Daniel: are there features that would be problematic for transpilers, linting, tree-shaking? Or beneficial
28+
- Refael: how can we simplify bundling? Standardized AST or intermediate representation? E.g., V8 team doesn’t want to expose its AST b/c they don’t want to worry about breaking people, so that’s not where we can get such a thing--would need to come out of standards.
29+
- Daniel:
30+
- Emerging set of standards that wants to reduce the need for bundling
31+
- “Webpackage”, “bundled exchange” is being prototyped in Chrome
32+
- A number of difft resources bundled into a single file, sent over the network
33+
- The hope is that HTTP/2 push would make it cheap to send many items, but browser support and configuration was bad
34+
- With bundled exfchange it woulde include html, css, js, images, video, in a single file
35+
- Refael on bundling:
36+
- In tooling world, we have different concerns
37+
- Bundling tools themselves may or may not be helpful
38+
- Npm is 27000 files itself when installed
39+
- Ben: webpack and bundlers create confusion for people
40+
- Moving towards a standard would reduce confusion in the community
41+
- Refael: why can’t I webpack my backend? It’s not applicable, but it’s confusing
42+
- Would ask TC39 to think about JS outside of the browser
43+
- Daniel:
44+
- Bundling is not TC39 work, it’s W3C stuff
45+
- “Import maps” would be necessary (?) for this to work. Web incubator community group, not at TC39 either
46+
- Ben: at a meta level there’s a disconnect between how a web browser and node represents files
47+
- Finding more common ground
48+
- How can we make URIs more palatable to the backend, for example
49+
- Nobody wants to provide a URI, they just want to require lodash
50+
- Daniel:
51+
- Import maps can solve this problem maybe
52+
- Chris: import maps? Not TC39?
53+
- Daniel: sometimes its unclear where standards should be developed
54+
Ben: what’s the best way to have dialog with TC39?
55+
Daniel: still trying to figure it out. Proposals are developed in open on GitHub, but this doesn’t always work because the people on the proposals don’t have the full context of the people giving feedback
56+
- Daniel wants to talk to them separately and bring feedback back to TC39
57+
- But may not be the best idea to have an intermediary
58+
- Ben: core node collabs have a node-centric view, there’s userland people using node, there’s browserland and browser vendors… nobody sees eye-to-eye necessarily
59+
- How can the group bridge the gaps?
60+
- Daniel: we want people prototyping stuff using emerging standards
61+
- Ben: like babel. We can reach out to tooling authors
62+
- Chris: does TC39 know about esm? (userland module)
63+
- Daniel: Yes
64+
- Daniel: experimentation like using webpackage would be helpful
65+
- Ben: c8 uses v8 coverage, which is experimentation
66+
- In the module discussion, it’s frustrating
67+
- We want to move forward our ecosystem
68+
- Package maps. But can’t move forward without experimental tooling
69+
- Daniel: Give feedback to proposals within github issues
70+
- Chris: a suggestion box!
71+
- Daniel: We thinking about setting up discourse
72+
- It can be hard to extract the feedback from angry people
73+
- Ben:
74+
- The impact of a tool can be more valuable than a discussion
75+
- More tools, less conversation!
76+
- Daniel:
77+
- There’s an import maps polyfill somewhere
78+
- Guy Bedford (not present) owns this
79+
Refael: Give us an idea of where TC39 sees itself in the ecosystem
80+
- Daniel:
81+
- TC39 sees itself as responsible for the language
82+
- But it’s unclear where some things land? Are they in the language?
83+
- Please participate in node open standards repo?
84+
- Have a call thereafter
85+
- Ben: Maybe regular TC39 rep, but not entire meeting
86+
- Daniel: Will try to show at next meeting to keep conversation going
87+
88+
- Chris: meeting how often? 3wks ok?
89+
- Ben: let’s slack more
90+
- Chris: maybe go to blessed slack
91+
- Rick: yes
92+
- Rimraf: Sam Roberts (IBM) will comment on rimraf-in-C++ issue
93+
- Ben: Node is now using my coverage thing
94+
- Sourcemaps are a big problem
95+
- Daniel: There’s lack of a proper standard. That’s a problem
96+
- Edge cases differ in certain places
97+
98+
## Announcements
99+
100+
\*Extracted from **tooling-agenda** labelled issues and pull requests from the **nodejs org** prior to the meeting.
101+
102+
## Q&A, Other
103+
104+
## Upcoming Meetings
105+
106+
- **Node.js Foundation Calendar**: https://nodejs.org/calendar
107+
108+
Click `+GoogleCalendar` at the bottom right to add to your own Google calendar.

0 commit comments

Comments
 (0)