Add an agenda for the 03-26 meeting. (#248)

And add the notes for the 03-12 meeting.

Author: sunfishcode

meetings/2020/WASI-03-12.md

@@ -35,3 +35,51 @@ Installation is required, see the calendar invite.
           the I/O story as we contemplate new kinds of streams and messages.
 
 ## Meeting Notes
+
+Attendees:
+
+Dan Gohman
+Jan Falkin
+Mike Trinkala
+Lee Campbell
+Andrew Brown
+Ralph Squillace
+Johnnie Birch
+Benjamin Brittain
+Mark McCaskey
+
+Meeting notes:
+
+Crypto API now has its own repository:
+https://github.com/WebAssembly/WASI-crypto
+Let's talk about process and expectations.
+
+Merging send/recv into write/read:
+https://github.com/WebAssembly/WASI/pull/240
+This is a complex proposal, but the intention is to simplify the I/O story as we contemplate new kinds of streams and messages.
+It’s difficult to comment because it’s a big and fairly abstract PR
+Prototype implementation would help people see how this works out in practice
+Impromptu questions:
+Is it ok to have ambient authorities for debugging facilities such as logging?
+BB, Lee: Yes, handles are good
+Making it ambient is a decrease in potential power of the system
+
+Impromptu agenda item:
+
+DG: Is it ok if we have ambient authorities producing handles if the handles themselves can’t be used to do any I/O?
+Sealed capabilities are a useful concept.
+Lee: Avoid ambient authorities.
+BB: Capabilities aren’t just security, they also allow compositional design patterns.
+Lee: eg. logging. Where does the log output go? If you make it a handle, now you can send different handles that represent different sinks.
+Logging for debugging vs logging for production, perhaps are not so different use cases.
+Lee: A really powerful feature would be to have move-only semantics for handles, so you can send an instance a handle and it can be sure it’s the only instance with access to it.
+DG: Reference types proposal doesn’t have move-only semantics, but it’s something to think about
+Lee: Reference-counting can lead to awkward edge cases where references are kept live, or have aliases, in surprising ways. 
+In Fuchsia, fd’s are 64-bit numbers and not reused.
+WebAssembly/interface-types
+BB: Move semantics changes how I’d design WASI APIs
+Eg. capabilities for memory regions. What does it look like to transfer a handle for one of these to another instance?
+Lee: Let’s have a discussion on rights, what rights look like, do we have a handle hierarchy?
+WASI doesn’t predefine __unix__.
+RS: GIve people a new engine, and it’s an easier step for people to take to port to. Guide them to doing the thing they want to do. Think of this as a new target. Don’t roll in old metaphors.
+Lee: If something doesn’t build, then you know what you have to fix. If it builds but doesn’t work, now you have to debug it.

meetings/2020/WASI-03-26.md

@@ -0,0 +1,51 @@
+![WASI logo](/WASI.png)
+
+## Agenda for the March 26 video call of WASI Subgroup
+
+- **Where**: zoom.us
+- **When**: March 26, 16:00-17:00 UTC
+- **Location**: *link on calendar invite*
+- **Contact**:
+    - Name: Dan Gohman
+    - Email: [email protected]
+
+### Registration
+
+None required if you've attended before. Email Dan Gohman to sign up if it's
+your first time. The meeting is open to CG members only.
+
+## Logistics
+
+The meeting will be on a zoom.us video conference.
+Installation is required, see the calendar invite.
+
+## Agenda items
+
+1. Opening, welcome and roll call
+    1. Please help add your name to the meeting notes.
+    1. Please help take notes.
+    1. Thanks!
+1. Proposals and discussions
+    1. WASI currently uses a POSIX-like `errno` enum, but that seems
+       not well suited for less POSIX-like APIs:
+       https://github.com/WebAssembly/WASI-crypto/pull/13#issuecomment-599324384
+       Many systems have evolved to having a single error enum shared
+       across the whole system. Is it ok if we don't aim for that?
+    1. POSIX tries hard not to let syscalls cause segfaults, returning
+       things like `EFAULT` instead. There may be some value in this for
+       POSIX compatibility, but how valuable is this otherwise? Should
+       other WASI APIs be allowed to trap?
+    1. Evolving idea: wasi-sdk API for extra emulation:
+       1. `-D_WASILIBC_EMULATED_MMAN`
+       1. `-D_WASILIBC_EMULATED_PAUSE`
+       1. `-D_WASILIBC_EMULATED_RAISE`
+       1. ...
+       1. `-D_WASILIBC_EMULATED_ALL` ?
+       and
+       1. `-lwasi-emulated-mman`
+       1. `-lwasi-emulated-pause`
+       1. `-lwasi-emulated-raise`
+       1. ...
+       1. `-lwasi-emulated-all` ?
+
+## Meeting Notes

meetings/README.md

@@ -26,3 +26,4 @@ Meetings of the WASI Subgroup of the W3C WebAssembly Community Group (CG) follow
  * [WASI January 16th video call](2020/WASI-01-16.md)
  * [WASI February 27th video call](2020/WASI-02-27.md)
  * [WASI March 12th video call](2020/WASI-03-12.md)
+ * [WASI March 26th video call](2020/WASI-03-26.md)