edgedb / website Goto Github PK

If I search for a dev/v3 feature, it gets a label in the search indicating it is a v3 feature as expected.

If I click through on it, the site loads the article, but it does not change the version selector to the correct version. As a result, the "New" tag next to the title is missing. This could suggest that the feature is actually part of the stable version since we're not indicating in any way that the feature isn't available in the currently selected version.

Clicking through on a dev/v3 feature from search should probably automatically update the version selector to first version that feature is in if the feature doesn't exist in the currently selected version. This way, it's more clear to a reader where that feature is available. This would be the ideal state after clicking through on the search:

In the EdgeDB 3.0 blog post, the docs link for splats links to localhost:3000 and not edgedb.com.

The current Rust documentation on the web site is a stub. Full docs are generated by the binding but are linked from the stub instead of integrated into the site.

The line

echo deb [signed-by=/usr/local/share/keyrings/edgedb-keyring.gpg]\

isn't right. The [ and ] need to be escaped, or the whole thing quoted, or something. Maybe there's some shell it'll work with, but for the bourne shells, it'll be turned into any of the one-character filenames whose names match one of the characters in "signed-by=/usr/local/share/keyrings/edgedb-keyring.gpg", if you have one. Otherwise it will fail.

This link:

:ref:`configuration parameter <ref_std_cfg>`

references the top of the config page but lands here:

which is most of the way down the page. I previously attempted to add another anchor to this same page further down the page and link to it, and it also landed at this same position when clicked.

This isn't isolated and seems to affect other links in the documentation as well.

@tailhook:

It's not specific to your change. If you click another reference it does the same. As I can see that's not leading to any specific section on the page, it just doesn't reset the the page to the top.

And I have to admit that all title links work like that. Just go to the changelog and click some links there.

I suspected this too: that the previous scroll position on a page was preserved when you returned to it via one of these links, but I don't think that's the issue here. I just tested with this link:

:ref:`with <ref_eql_with>`

I clicked it which lands me here:

about 1/3 down the page even though the anchor is at the top of the "With" page. I scrolled to the top of that page, hit back to return to the page containing the link, and clicked it again. It again landed me at the same position.

In float64's data wire documentation, the link goes to std::float32 and not std::float64. Other pieces of information for float64 is wrong as well. I think it was copied from float32 without the information being changed.

Allow the user to limit searches in our documentation to a particular category. Among other things, this would allow for users looking for help with a particular client to limit their query to that client's documentation.

I found a few while reading:

• https://www.edgedb.com/tutorial/building-blocks/scalars/cal - "duraitons" instead of "durations". Also "al" instead of "all" on this same page.
• https://www.edgedb.com/tutorial/basic-queries/config/by-expression - unnecessary "the" at ...is that the it should be possible...
• https://www.edgedb.com/tutorial/data-mutations/insert/using-for - unnecessary "the" in a comment at ...and the unpack it...
• https://www.edgedb.com/tutorial/data-mutations/delete/deleting-data - "an an expression" instead of "as an expression"
  Also an unnecessary comma on this same page at ...since it has an, exclusive constraint on it...

I would like to propose adding a showcase page to the website. This extends an idea that was discussed between moderators of the Discord. I'd love to hear more feedback on this, I think this has good potential for bringing more awareness. Here's a rough list of reasons why:

• You have to use #showcase currently to find communal examples or live projects.
• Some repositories have an edgedb tag, some do not. This makes it hard to find all public examples.
• There isn't a location where non-Discord and non-GitHub users can quickly view what projects exist using the service.

Here's some ideas I have for making this possible:

• We could maybe automate the page to showcase GitHub repositories.
  • Sometimes the edgedb tag is used, this could be referenced.
  • GitHub can intelligently recognize used packages through a package manager file, i.e. requirements.txt, Cargo.toml and etc.
    • Maybe scraping for these would help to give a better list of projects, including those missing the edgedb tag.
    • Client libraries could potentially add a setting for projects to be seen for analytical/non-commercial purposes.
• (Alternative idea suggested by @raddevon) We could find a way to maintain or curate the list.
  • I think most public examples of EdgeDB being used will be posted on GitHub. Maybe we could hand pick some projects that are known to work

Seeing from setup.sh that the documentation for other libraries are being built from their repositories, I would like to propose using git submodules instead, like the studio path in this repository. It would allow finer control over synced repository changes for documentation, and since GitHub will recognise the latest commit, you can always pin to a certain commit hash, or keep it up-to-date.