Links to sample data cause a 404

[philrz]

A community user reported the following via private gist.

There's links to sample data from the docs that currently result in a 404. For instance, the one linked from the section https://zed.brimdata.io/docs/next/tutorials/zed/#adding-data-to-our-lake highlighted here:

When clicked:

It doesn't fail when clicked in the page's original location within section https://github.com/brimdata/zed/blob/main/docs/tutorials/zed.md#adding-data-to-our-lake on GitHub, however. The original markdown link is just [here](github1.zng), and when clicked from within GitHub that resolves to https://github.com/brimdata/zed/blob/main/docs/tutorials/github1.zng. For what it's worth, that may not be a great link destination either, because it just brings the user to a page at https://github.com/brimdata/zed/blob/main/docs/tutorials/github1.zng that looks like this:

At this point clicking either "View Raw" or that "Download" button both end up downloading the file. So having this link resolve to the intermediate screen doesn't seem to be helping the user much.

Meanwhile, on the deployed docs website, the hyperlink that 404s ends up as https://zed.brimdata.io/assets/files/github1-2496d524ce44f4cb9f32cfeb70fa669e.zng/. Based on what I see in the Docusaurus docs, I guess it recognizes it's a file destination and downloads it and serves it out of this assets/ directory. In any case, as it turns out, if we just remove the trailing slash, the link works fine as https://zed.brimdata.io/assets/files/github1-2496d524ce44f4cb9f32cfeb70fa669e.zng.

Therefore, I can see see at least two possible ways out, though I'm sure there's others:

1. We could change the destination for the link in the original markdown to be https://github.com/brimdata/zed/raw/main/docs/tutorials/github1.zng (i.e., the link attached to that "Download" button), since it looks like that link would stay as-is even after Docusaurus publishes the site. This would have the added bonus that if users ever click the link in the original page on GitHub it'll go straight to download, but maybe there's other negatives to having an absolute URL like that.

2. Someone could figure out how to remove the trailing slash from the generated link on the docs site or figure out how to make our web site serve a file destination even if it has a trailing slash. FWIW, I did some searching and found people at asset links with trailing slash true can be broken facebook/docusaurus#6282 talking about a similar problem.

[philrz]

I've reopened this. Details are in #32.

[philrz]

The conclusion here is that we ended up using a Docusaurus config with trailingSlash: false and hosting the site at Netlify rather than GitHub Pages to make that config workable. The links to the assets on the Zed tutorial page that previously caused an HTTP 404 when clicked now have links https://zed.brimdata.io/assets/files/github1-2496d524ce44f4cb9f32cfeb70fa669e.zng and https://zed.brimdata.io/assets/files/github2-d64f82ea518d971a86ad9e81da3a9b6e.zng, which download successfully.