Propagate FileReader Errors about rosbag.js

[janpaul123]

Interesting! I don't think we've loaded such big bag files in memory — usually we'd read them off the filesystem directly when they're that big. Any chance you might be hitting some memory limits? Maybe the bit about permissions is a red herring (especially since they say "typically due to")?

[gkjohnson]

Yeah I'd expect that downloading a 1GB file blob would fall into the "not recommended" category -- we've been using smaller bags for our work until now.

... usually we'd read them off the filesystem directly when they're that big.

We're considering building out an Electron app to support cases where our bag data is so large. Is that how you're reading files directly off the file system? Or are there other ways you're using to read files directly in the browser (such as the file input field)?

Maybe the bit about permissions is a red herring (especially since they say "typically due to")?

The error message definitely struck me as a generic one so I'd believe this.

Thanks!

[janpaul123]

We're mostly using the drag and drop api. It gives you a File object with the only difference being that the data remains on disk. You can pass it into open from rosbag just like above.

[janpaul123]

And yeah a file input field works the same way.

[jtbandes]

Regardless of the cause, you're right that it seems to make a lot of sense to propagate the onerror event from the reader. As for reproducing it for testing, it does seem surprisingly tricky to force a failure, but it would work to instead just mock out the readAsArrayBuffer function for testing so that it just asynchronously calls the onerror.

[Katerhong]

so, What is the conclusion?
I feel confused

[jtbandes]

The original issue was fixed and published in version 1.3.2. If you are having trouble, I would recommend filing a new issue. By the way, you might also want to check out @foxglove/rosbag, a fork of this project with some updates to the API and more bug fixes.