Week 36 - 2019
9/2 - 9/8
- Implemented Response.statusText. This was necessary for my book to explain the fetch response object.
deno versioncommand used to use V8 Isolate for showing the version number. This PR changes it. Now
deno versionis only processed by the rust side. It's partly done by @nayeemrmn.
- We used to use rollup-plugin-replace for replacing arbitrary symbols in TypeScript source code. That was handy technique, but we recently ditched rollup and now that string replacing behavior was very naively performed in our source code. This PR removes that hack and simply passes everything on start up message from rust to V8. The extension of such message has become really easy because of the recent ops infrastructure change. The messages are now simply JSON. So what I needed to do was just adding fields to a JSON object.
- Response.bodyUsed wasn't implemented. I tried it, and it wasn't difficult. I had a little concern about the timing when the bodyUsed becomes true, but when I checked it in Google Chrome, the browser seemed making it true immediately after the response.text() called. So I mimicked that behavior in the above PR.
A big change in build pipeline happened this week. Rollup was removed. We now load scripts one by one and take the snapshot of the entire state. By doing this, we don't need any script bundler! This was a big surprise to me. I've never imagined that such a way even existed.
And then we removed
//tools/build.py. This was a big step forward because now we build with standard
cargo buildcommand. This means anybody now can add any crate as a dependency. The rust side of deno is now truly open for everyone to extend.
- This week lambda started releasing the update which speed up the cold start up time of lambda functions which connect to VPC. It used to be very very slow because of the start up time of ENI was very very slow. Now AWS provides new infrastructure for ENI initialization. So the lambda is now very quick to start up 👍
- According to the above article. Google Chrome is to change the browser's
default behavior around
SameSite=attribute of Set-Cookie header, and the article says it will result the end of CSRF attacks. I've never checked the above assertion in detail, but if it's true, we can simplify the typical web development process significantly.
- According to the above article. Google Chrome is to change the browser's default behavior around
🦕 denobook 02 is the 2nd volume of denobook by deno-ja
- I finished the editing of the book.
- Have written many things using gRPC.
See any mistakes? Please fix!