Closed finnbear closed 4 years ago
Well, yes. When using my search engine of choice, it looks as expected:
When clicking on the search result, you directly land on the Matrix4
page.
Sorry, I clearly don't vote to make the proposed change and break all existing URLs just because search engines can't do their job right.
Does DuckDuckGo support hashbangs, or can we not use the URL fragment at all?
This isn't necessarily a large change, it might just mean adding a simple redirect.
This isn't necessarily a large change, it might just mean adding a simple redirect.
How would this look like? Can you configure redirects with GitHub pages?
It used to be the case that #!
was recommended with Google Search, and #
basically wasn't supported. Redirecting from one to the other is just a JavaScript change.
Now it doesn't matter with Google Search. I don't know about DuckDuckGo. But you're right, if the hashbang thing doesn't work and we need to use real /docs/foo
paths, we can't do this on GitHub Pages.
Something I probably should have mentioned is that you already have URLs without the '#' that can be found with a slightly longer search term.
However, their title of "[name]" makes them difficult to see at first glance.
If you just fixed these pages to have the proper title, maybe duck duck go and the other search engines I mentioned would index them more readily.
This is a related issue that may explain the mentioned pages: https://github.com/mrdoob/three.js/issues/10937
So their crawler doesn't support javascript?
I can't speak to what DuckDuckGo or any other engine does internally, and I'm not suggesting you break any links or do the impossible. It seems like it would be possible to fix the missing [name]. https://github.com/mrdoob/three.js/blob/3c13d929f8d9a02c89f010a487e73ff0e57437c4/docs/api/en/math/Matrix4.html#L11
@mrdoob If you would be open to a PR, I would be willing to go through and add a
Have you considered to report this issue to DuckDuckGo first? Fixing the issue on their side seems more appropriate. It then works for all websites using the current docs approach.
@mrdoob If you would be open to a PR, I would be willing to go through and add a
to each html file
Sure! As long as you are willing to continue updating all the relevant files for the rest of your life 😁
More seriously, this is an issue with DuckDuckGo's crawler.
What was I thinking? There is a much easier, standards-compliant way to fix this issue: adding a sitemap. Please review #19037 :smiley:
(And yes, I contacted DuckDuckGo. Their automated system said they had a high volume of messages and would take time to respond. And this is a problem with the other search engines I mentioned, not just DuckDuckGo)
It's the same on Bing:
Maybe DDG uses Bing?
Maybe DDG uses Bing?
Well, technically yes.
DuckDuckGo's results are a compilation of "over 400" sources, including Yahoo! Search BOSS, Wolfram Alpha, Bing, Yandex, its own web crawler (the DuckDuckBot) and others.
From: https://en.wikipedia.org/wiki/DuckDuckGo
I don't think either Bing or DuckDuckGo is to blame though. By using google analytics, you may be giving Google an advantage when it comes to indexing the page as the user sees it :wink:
Having a proper sitemap is a good first step for any search engine, hence the PR. By having the sitemap point to the urls with the '#', and thus the proper title, instead of the raw html pages, the [name] issue may be fixed.
In the unlikely event that it doesn't fix the [name] issue, some sort of
More seriously, this is an issue with DuckDuckGo's crawler.
Yes, I don't think a PR makes sense in this context. Hence, a sitemap is also not necessary.
Again, this needs to be fixed on search-engine website.
Description of the problem
When I want to look at the docs for a particular Three.js class, I type 'three.js [classname]` into my search engine of choice (DuckDuckGo). The only relevant result is the docs, but not the specific docs page. I then have to click the docs and search for the class again.
Feature request
If possible, facilitate indexing of all three.js docs pages. This may mean switching from '/docs/#api/etc' to '/docs/api/etc' urls, which could replace or forward to the existing pages.
Search engine