Hi, me and my group are taking a crack at this as part of a uni course for the next week. Could we get assigned this?

@Glassar It is yours! :-) If too hard, easier issues like #5565 or #4895 are available.

(note for GSoC: If this bug gets solved, we will replace the GSoC's tasks with another of similar size)

Thank you, and great to have those suggestions. We have done a first code inspection and think this should be doable though so we'll start here

We have done quite a bit of digging and found where an upload attempt fails (in uploadWorker, or this is at least where the failure is determined).

So our plan right now is to add code to the "failure" points in UploadWorker.uploadContribution which passes the failure reason onwards. This can then be used in ContributionsFragment.retryUpload to determine if this is a genuine failure and the attempt abandoned, or if it should try again. Finally this failure reason could also be used give the user better information about why their upload has failed repeatedly link.

But to determine what is a genuine failure, we need to know/decide what is a genuine failure which can the be used as a whitelist to stop retrying (one obvious example which won't change is if we get a FileNotFoundException). And this might be identified through for example exception messages, but we are having some trouble with determining what those exact conditions might be. Any ideas of the kind of errors might belong here, or where we should look?

Another thing we are wondering about is passing the information. The easiest and most intuitive way would be to add another field to the Contribution class for an error message as this how if it failed is already passed around. But the contributions are stored on the DB and as such we are not sure it storing the failure information there is optimal, any tips here?

Finally it seems reasonable to notify the user about what is going wrong with their upload, both when it fails to many retries and when we determine that it is a genuine failure. In what way should these notifications be sent (as far as I can see there is already sent something to the user in UploadWorker when it fails)?

One case I sometimes see is when the picture has been uploaded already, but the app somehow believes it failed, and retry.

I shared one reason for the case Nicolas mentioned above in this issue: #5280, might help.

Any ideas of the kind of errors might belong here, or where we should look?

Invalid filenames also cause this issue. I tried adding a § sign in the caption, it keeps failing.

Any ideas of the kind of errors might belong here, or where we should look?

Invalid filenames also cause this issue. I tried adding a § sign in the caption, it keeps failing.

Thank you, its super useful to have a way to reproduce this issue. Do you know of any other patterns that causes it to fail?

Invalid filenames also cause this issue. I tried adding a § sign in the caption, it keeps failing.

Hey, I was just trying to reproduce this issue, but surprisingly for me, the image successfully uploaded with a § in the caption:

We have gotten quite far on the way to wrapping this up and it should techinically be able to be roled our (more details in the PR), but we are unable to continue due to our course running out.

The things left to implement is a more complete whitelist over "genuine" failures as we have not been able to actually reproduce any examples of it. Also it would be good if the user was notified about said "genuine" failures and so they can address it.

The caption has to be only § for the upload to fail. A caption with § at the end does not fail.