"v3" loads wrong JS

[bkroeker]

<%= filestack_js_include_tag %>

loads //static.filestackapi.com/filestack-js/1.x.x/filestack.min.js

but it should, I expect, load //static.filestackapi.com/filestack-js/3.x.x/filestack.min.js

My code wasn't working until I manually loaded the latter URL and bypassed the filestack_js_include_tag method.

[gabifija]

Hi @bkroeker ! I have updated filestack-rails to version 5.0.0. By default we setup filestack-js version to 3.x.x. You can also provide a version which suits you best. If you want to change the version, you have to add specific version in configuration. Here is an example: https://github.com/filestack/filestack-rails#filestack-picker-version

[bkroeker]

@gabifiolek Thanks! Seems to work well.

[bkroeker]

@gabifiolek One thing I'm noticing is that a bunch of my rails tests began failing. I tracked it down to Picker#url_exists? which was swallowing the VCR error that was being thrown. The fix, of course, is that I need to enable vcr on some additional specs, but it would have been nice to have a more accurate error message than "Incorrect config.filestack_rails.version". Maybe consider scoping the rescue to only catch URI::InvalidURIError or something?

That said, I'm not particularly thrilled about my site making a request to check if "3.x.x" is valid on every single page load. I'd much prefer if the filestack-rails gem just ran it against a regex or something.

[gabifija]

@bkroeker Thank you for your comment!

I was thinking of a regex to check if the version is correct. But, the problem is that a regex has to be updated each time when we release filestack-js version. For instance: when we update filestack-js to newer version - 5.2.1, regex will fail. I can not create a regex for random digits, because file url (https://static.filestackapi.com/filestack-js/#{version}/filestack.min.js) doesn't not exist for random version. URL works only for specific versions. That's why I have to check if the URL exists, and if we have an access to it.

Does it make sense?

What I can do in that case - I can move checking if url_exists? to configuration. It will not check it every single page load. https://github.com/filestack/filestack-rails/pull/219/files ?

Let me know if I can be any further assistance.

[gabifija]

@bkroeker FYI, I moved url validation to configuration in filestack-rails 5.1.0. Let me know if that works for you better. Any suggestions always welcome.

[bkroeker]

@gabifiolek Yeah, I totally get the reason why you went that route rather than a regex, but IMO it's coding too defensively and the hidden side effects aren't worth it. You're holding developers' hands more than is warranted.

Scenario 1: Let's say you validated via a regex. What would happen if the user typed in 3.9.9 and that version didn't exist? Their page would list a 404 that //static.filestackapi.com/filestack-js/3.9.9/filestack.min.js could not be found. They'd say "Oh, why doesn't that file exist?", they'd look it up, and find a version that does exist.

In this case, the regex would be just a sanity check to make sure they're not typing in something totally bogus. A bit more responsibility would fall on the user to make sure their version is valid, and I think that's fine. You could put a link on the readme to a place where they can look up the available versions.

Scenario 2: let's say you kept the current code, and there's a new major version of filestack released: 4.x.x, and it has some backwards-incompatible changes to its api. Your current code that checks for its existence would pass, but the gem could break despite its attempts to be foolproof.

With a regex like /[123]\.(x|\d+)\.(x|\d+)/, you could control for major versions. 4.x.x would not be considered valid until you've tested it, fixed any broken behaviour, and updated the regex. Once the regex is updated, everyone would know that it's been tested and is ready for 4.x.x. This is a safer and better solution, in my opinion.

Current Scenario: Now, to the changes you just made for 5.1.0, I don't think the behaviour actually changed despite your changes to code. Ultimately, the url validation code is still triggered by the call to filestack_js_include_tag, which means it's triggering on every single page load. I will continue to avoid calling filestack_js_include_tag and load the js manually to avoid this behaviour.

Thanks for listening.

[gabifija]

@bkroeker Thank you for giving me a feedback! I really appreciate it.

I've talked with our team and we decided to remove version validation (in 5.2.0) I left a comment in README to check available filestack-js versions.