fix: Ensure that each request can only fail once. #28
Conversation
|
I think there may be another bug where you call close while a reconnect is already scheduled, but I will handle that in another PR. |
| 'ms, assuming connection is dead' }) | ||
| // Timeout doesn't mean that the request is cancelled, just that it has elapsed the timeout. | ||
| destroyRequest() |
There was a problem hiding this comment.
Being more careful about destroying requests will help to ensure applications are not held open by the event source.
|
|
||
| module.exports = g; | ||
| var g; | ||
|
|
There was a problem hiding this comment.
this just formatting or was there a change?
There was a problem hiding this comment.
This file is auto-generated based on the other files. I am not sure the cause of the diff here. It looks like most of the formatting is even the same.
| * Wrap a callback to ensure it can only be called once. | ||
| */ | ||
| function once(cb) { | ||
| let called = false |
There was a problem hiding this comment.
It sure is nice to be single threaded
There was a problem hiding this comment.
It is very nice indeed. (at least the illusion of a single thread.)
| @@ -12056,4 +12080,4 @@ module.exports = function isBuffer(arg) { | |||
| } | |||
|
|
|||
| /***/ }) | |||
| /******/ ]); | |||
| /******/ ]); | |||
There was a problem hiding this comment.
I'm not sure if it was manually edited before, but it generated this way.
| if(!called) { | ||
| called = true | ||
| cb(...params) | ||
| } |
There was a problem hiding this comment.
Is there a logger util in this library? It's helpful to see logs when something is ignored.
🤖 I have created a release *beep* *boop* --- ## [2.0.3](v2.0.2...2.0.3) (2024-05-22) ### Bug Fixes * Ensure that each request can only fail once. ([#28](#28)) ([bcceb35](bcceb35)) * Stop reconnect timer when event source is closed. ([#29](#29)) ([a73a118](a73a118)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
On WSL with no network, or running docker without a network, the 'error' and 'timeout' callbacks would both be called. The timeout should only happen after the specified timeout, 5 minutes in the node SDK, but happens immediately under either of those conditions.
This change ensures that each request can only fail one time.