|
1 | 1 | # oak Change Log |
2 | 2 |
|
| 3 | +## Version 17.0.0 |
| 4 | + |
| 5 | +Releasing as a major release out of the abundance of caution in that the ability |
| 6 | +to re-read the body request is a breaking change (in a good way), but the update |
| 7 | +to oak/commons v1 and deno/std v1 introduced a couple of knock on minor breaking |
| 8 | +changes that might change some behaviors in unexpected ways. |
| 9 | + |
| 10 | +There were no changes to support Deno v2 release candidate, but this is the |
| 11 | +first release to be tested against Deno v2. |
| 12 | + |
| 13 | +- feat: ability to re-read the body of the request (e099557) |
| 14 | + |
| 15 | + When the body API was re-aligned to the Fetch Request API in v13, it also |
| 16 | + followed that the body can only be consumed a single time. This is unexpected |
| 17 | + for those using a middleware framework, where people generally expect to be |
| 18 | + able to read the body several times as they process the request. |
| 19 | + |
| 20 | + This feature memoizes the body based on the type it was originally read as and |
| 21 | + then can be subsequently read as a compatible type. |
| 22 | + |
| 23 | + If you read it first as one type and then switch to an incompatible type, the |
| 24 | + method will throw (for example, reading as an array buffer but then attempting |
| 25 | + to read as text). |
| 26 | + |
| 27 | +- refactor: use PropertyKey instead of key list (#662) |
| 28 | +- chore: run CI on canary (#672) |
| 29 | +- chore: migrate to std and commons v1 (d6f9d38) |
| 30 | + |
| 31 | + There were some minor behavioral changes with media-types and HTTP errors that |
| 32 | + have impacted the behavior of oak. |
| 33 | + |
3 | 34 | ## Version 16.1.0 |
4 | 35 |
|
5 | 36 | - feat: support authority pseudo header for request (0c8b8da) |
|
0 commit comments