Remove right margin from custom scrollbar

[rotemdan]

Here's a basic pull request for an attempt at fixing #1492

You can use it as a starting point. It's not tested (tried to run npm run preview locally but it failed due to lack of MongoDB, seems like getting this to work isn't trivial for me and it'd take some more time and effort than I have right now).

I simply removed the tailwind class mr-1, which means:

.mr-1 {
  margin-right: .25rem;
}

From instances of custom-scrollbar in these source files:

src/lib/components/chat/ChatWindow.svelte
src/routes/assistants/+page.svelte
src/routes/models/+page.svelte
src/routes/tools/+page.svelte

You'll need to test it yourself to ensure it actually works.

[rotemdan]

I've managed to start it locally using a docker container for the MongoDB instance.

The local instance doesn't fully work (getting errors like Authorization header is correct, but the token seems invalid) but I could paste a large amount of text to the chat input to see how the scrollbars behave.

Based on what I've seen, it seems the change does apply correctly.

Just note that Chromium has an unrelated issue with scrollbars: on my PC, for a zoom level of 125%, the scrollbar isn't pressable at the right edge, but with other zoom levels (100%, 150%, etc.) it does work. I've seen this happen on many other websites. It could be some internal issue with rounding of fractional pixels, or something of that sort. I'm not sure if it's actually reported on the Chromium tracker.

[nsarrazin]

Works great! Thanks for taking the time to create a PR 🤗

[rotemdan]

Issue has now returned. I'm not sure exactly when, but I think even yesterday it still worked correctly.

Looking at the updated source code, the files, where I applied changes to remove the mr-1 classes from tags, have changed significantly.

I see the mr-1 and mr-1.5 classes added in many different places. The code looks very different from when I made the changes previously.

Figuring out what changes to make to fix the issue was hard enough the last time. I don't know if I could go through it again. As a maintainer of many different open-source repositories myself, I'd never ask a user for a second pull request on an issue they correctly fixed earlier, if the reason it broke again is not related to them.

I reopened #1492.

As you can see, I personally really care about this particular behavior. I noticed it very quickly, and it's frustrating to have it back again.

[rotemdan]

Turns out the return of the issue was actually a Chromium bug I activated by accidentally changing to a 125% zoom level. It wasn't coming from the website. I closed the issue again.