Using http-semantics #114
Using http-semantics #114
Conversation
|
@kazu-yamamoto I had to make a few small adjustments, but only minor, and tests pass just fine. Two of the changes I had to make were because there two places where I could have imported from One small comment is that There is a downside to re-exporting entire modules (as in re-exporting |
|
Thank you for your review. |
In kazu-yamamoto#92 we added an exception handler that was meant to catch _all_ exceptions (sync and async). This got changed in kazu-yamamoto#114 (specifically, kazu-yamamoto@52a9619): when we moved from `Control.Exception` to `UnliftIO.Exception`, we got a different behaviour for `catch` and friends (see well-typed/grapesy#193 (comment)) for a full list. This commit fixes some unintended consequences of this change.
In kazu-yamamoto#92 we added an exception handler that was meant to catch _all_ exceptions (sync and async). This got changed in kazu-yamamoto#114 (specifically, kazu-yamamoto@52a9619): when we moved from `Control.Exception` to `UnliftIO.Exception`, we got a different behaviour for `catch` and friends (see well-typed/grapesy#193 (comment)) for a full list. This commit fixes some unintended consequences of this change.
In kazu-yamamoto#92 we added an exception handler that was meant to catch _all_ exceptions (sync and async). This got changed in kazu-yamamoto#114 (specifically, kazu-yamamoto@52a9619): when we moved from `Control.Exception` to `UnliftIO.Exception`, we got a different behaviour for `catch` and friends (see well-typed/grapesy#193 (comment)) for a full list. This commit fixes some unintended consequences of this change.
In kazu-yamamoto#92 we added an exception handler that was meant to catch _all_ exceptions (sync and async). This got changed in kazu-yamamoto#114 (specifically, kazu-yamamoto@52a9619): when we moved from `Control.Exception` to `UnliftIO.Exception`, we got a different behaviour for `catch` and friends (see well-typed/grapesy#193 (comment)) for a full list. This commit fixes some unintended consequences of this change.
This implements #113.
Backward compatibility will lost for some functions of HPACK but I guess it's a minor issue.
warpcan be built with this branch without modification.@edsko Could you try to use this branch? If this breaks your projects, I should provide more deprecated functions or types.