[🐛] Image compression is not consistently applied

[tao-qian]

[🐛] Image compression is not consistently applied

[tao-qian]

Verified on our end that image compression is now happening consistently! Thanks!

[khushal87]

Hey @tao-qian, thanks for bringing up this issue with us. So you mean this issue only happens when offline support is enabled by you?

[tao-qian]

Hey @tao-qian, thanks for bringing up this issue with us. So you mean this issue only happens when offline support is enabled by you?

I haven't actually tested this without offline support, and I just assumed that the behavior there would be different. IIUC, without offline support, the image will be in an "uploading" state with a spinner on top, once they are added to the input box, and the "Send" button won't actually be enabled, until the upload finishes. I assumed the "uploading" state would allow the compression to finish properly.

Whereas when offline support is enabled, the send button can be pressed right away as soon as the image is added. And based on my observation, if it is pressed without waiting for ~ 5 seconds, then the image that's sent won't be compressed.

[khushal87]

Thanks for pointing this out to us @tao-qian. This indeed looks like an issue. We will investigate and get this fixed and reach out to you soon. Thanks 😄

[tao-qian]

Thanks for pointing this out to us @tao-qian. This indeed looks like an issue. We will investigate and get this fixed and reach out to you soon. Thanks 😄

Thanks! Please keep us posted!

[khushal87]

Hey @tao-qian, we have fixed it in #2331 and its merged as well. Please wait for the next release and this should fix your issue. Do let us know if it doesn't work. Thanks 😄

[tao-qian]

Hey @tao-qian, we have fixed it in #2331 and its merged as well. Please wait for the next release and this should fix your issue. Do let us know if it doesn't work. Thanks 😄

Thanks a lot for the quick turnaround! We will keep an eye out for the next release then try this out!

[tao-qian]

We are seeing additional cases of images not getting compressed :(

This time it is harder to reproduce and I haven't figure out the consistent repro steps yet. However, it typically happens after I sent a few images in a row (e.g. 10). The worrying part for this bug is that once it happens, the subsequent images also won't be compressed. However, this sometimes would reset, after a text message is sent.

Here are two image messages for the same image. One of them is compressed as expected, but the other one is not:

• https://dashboard.getstream.io/app/1266909/chat/explorer?path=channels%2Cchannel~messaging%3Ahr_brightgo_254%2Cmessages%2Cmessage~254-781042d6-2b9d-4742-32f8-4257f070af0a
• https://dashboard.getstream.io/app/1266909/chat/explorer?path=channels%2Cchannel~messaging%3Ahr_brightgo_254%2Cmessages%2Cmessage~254-4240d9e7-365d-454c-0da5-aad937d06f25

Note that the two images practically look the same, but if you download them, one of them would be 400 kb while the other would be 1.5 mb.

We are overriding the compressImage handler to use the latest version of the ImageResizer library (@bam.tech/react-native-image-resizer). Here is our logic:

 compressImage: async ({
      compressImageQuality = 1,
      height,
      uri,
      width,
    }) => {
      try {
        const { uri: compressedUri } = await ImageResizer.createResizedImage(
          uri,
          width,
          height,
          'JPEG',
          Math.min(Math.max(0, compressImageQuality), 1) * 100,
          0,
          undefined,
          false,
          { mode: 'cover' },
        );
        return compressedUri;
      } catch (error) {
        Sentry.captureException(error);
        return uri;
      }
    },

We have confirmed that the input and output for compressImage didn't change when this bug happens:

• input parameters height, width, compressImageQuality, uri are all correct
• the uri in the output has the correct size (e.g. 400KB), when I checked it with react-native-fs

We are still trying to figure out more consistent repro steps, but just want to share the findings so far in case the behavior rings a bell. Thanks!