News

This is a list of project announcements such as new features sorted in chronological order, from the newest to oldest.

Significant entries will have a corresponding announcement on the following official accounts:

You can now update your profile picture on OurBigBook Web by uploading an image to the website like in a normal website.

Previously, we only supported linking to an external image URL. Now this is not allowed anymore and you must instead upload your image to the website. Existing external links will continue to work, but if you want to update the profile picture again, then you will need to upload your own next time.

Besides being a basic feature expected from any modern website, this is the first instance of "static file upload" on the site, and serves as part of a more general static file upload mechanism that can be later reused for other important features like uploading images for your articles.

This initial implementation is very simplistic: we are just storing the image directly in the database. We will look into migrating to a more proper static file solution later on if this starts to hurt performance. We're using the sharp Node.js image processing library, a frontend to libvips, to downsize input images as needed.

The \H disambiguate argument now shows on most navigation locations including:This makes it easier to understand what the topic is about when disambiguate is fundamental.

Announcements:

For some time we have been listing certain "cross article" metadata at the bottom of articles on both -W, --web and static.

For example, on both:you can see a list of articles tagged by the given articles at the end of the page.

Now, only on -W, --web, you can also see these article lists with the article content itself, for example:

The initial motivation for this was to be able to quickly browse through tagged articles, especially since the recent tagged headers show under non-toplevel headers.

Another motivation for this is the ability to be able to view such lists with pagination when a large number of items exists. While we don't currently limit tagged and incoming links listings, children listings are already useful as we currently limit dynamic article tree ToCs to 1000 entries, so that children listings open up a way to explore such large article trees.

This is the type of cute thing that can only be done efficiently on -W, --web, where we can use an actual database to build up a precise response as requested. On static websites, this would either require lots of repetition on pre-rendered HTML, or making several JavaScript requests to fetch individual articles from the server, which could risk overloading the server.

Announcements:

It is now possible to search through ToCs on static websites renders. Only items under the current ToC are searched for.

Ideally we should also implement a mechanism to search across all headers from any page. This would require generating a JSON with all the headers and their renders and fetching it from JavaScript. Doable but more work. For now we keep it simple and only search the existing ToCs.

Announcements:

It is now possible to search articles and topics by a given ID prefix.

We are going to look full text search for non prefix searches next. But at least the frontend is working and any improvements will be backend only.

Simultaneous search and sort (e.g. sorting by date and score) is also not available at present, once you start searching it disables sorting. We will also look into how to implement that.

Related issue: github.com/ourbigbook/ourbigbook/issues/263

Announcements:

Previously, tagged articles would only show up at the bottom of the page for the toplevel header, e.g. you'd only be able to see which articles are tagged as:under the split header page:but not under the non-split:Now tagged articles are also shown under non-toplevel headers, both on static websites and OurBigBook Web.

This makes it much easier for readers to view the tags, as it does not require them to click the header to view it as the toplevel and then go to the bottom of the page.

It also means that headers that are used primarily as tags and which would be otherwise empty now show some meaningful content.

Announcements:

OurBigBook CLI now forces you to write a consistent tree of headers by default, including e.g.:

It is a common issue with most plaintext note taking systems that they don't force you to make a consistent tree.

However, for publishing, having one tree is essential, otherwise it can be very hard for users to navigate your content.

Furthermore, this makes things much simpler to implement and understand for OurBigBook Web and a future WYSIWYG local editor.

Some other pedantry:

While OurBigBook Web topic mind melding remains the most innovative feature of the project, local plaintext is a fundamental guarantee that you will never lose your content, and we intend to keep it awesome.

Announcements:

At github.com/ourbigbook/ourbigbook/commit/075872a0a5ca7faf171d45834bc2b47995a15634 and nearby previous commits we've optimized the database queries made on article pages, mostly by adding some key missing indices and cache columns.

As a result, ourbigbook.com/cirosantilli now starts downloading the first byte 2x as fast as before, going down from about 1200 ms to around 600 ms, at a time region which makes a huge difference for user experience.

We will also start keeping better performance logs at: Section "OurBigBook Web performance log" to make sure we don't regress as easily.

Announcements:

We've greatly improved the Visual Studio Code extension adding support for the most important VS Code language features: Ctrl + T header search, Ctrl + click jump to header, header outline and link autocomplete

Thanks to Juhani Junkala for the awesome CC0 chiptune game soundtrack! opengameart.org/content/5-chiptunes-action

Announcements:

The article body now shows by default on all article lists. So do comment lists.

The major application of this is to quickly browser through a users's top or latest posts, e.g. ourbigbook.com/go/user/cirosantilli/articles?sort=score

Previously, the body would only show on:

Now it shows everywhere else as well, except that in other views, only a fixed height preview is shown to allow quickly going through large articles without too much scrolling.

A "view more" button can uncover the hidden content if the user wishes to usee it.

A "Show body" control was also added to toggle body vs the previously existing table mode.

Announcements:

It is now possible to mark articles as unlisted on OurBigBook Web: Section "OurBigBook Web unlisted articles".

The most important effect of this is that unlisted articles don't show on the table of contents of its ancestors. They also don't show on many article listing by default, e.g. on the list of user's latest articles.

The main use case we have for this feature right now is to stop polluting the table of contents with articles a user does not wish to show, and especially when doing local to Web upload, where Web articles are marked as unlisted by default if they are deleted locally.

We offer unlisted as an alternative to deletion for now because of the general philosophy what "permalinks should never break". This is currently not true as we don't have article history and therefore no permalinks. However, once history is implemented, we want to make it so links to specific versions will never ever break by forbidding article and history deletion entirely. Marking articles as unlisted will then allows to prevent deletion, while still keeping table of contents tidy.

In order to give more immediate topic value to readers, and to better highlight the topics feature, we now show a few articles on the same topic at the bottom of every article page, essentially acting as a preview of the corresponding topic page.

For example, if you visit the "Calculus" article by user Barack Obama: ourbigbook.com/barack-obama/calculus then at the bottom of the page you can see a section "Articles by others on the same topic (3)" which displays up to the 5 most highly upvoted articles in the same topic written by other users, much like the topic page for the "Calculus" topic: ourbigbook.com/go/topic/calculus.

By comparison, the topic page shows more articles by default (20), supports pagination, and allows for other forms of sorting such as viewing the latest articles in a topic. We are initially not adding those options to the article page itself as there is already enough stuff going on there.

Announcements:

Previously, when clicking a link to an element that is present in the current page, the URL fragment would contain the full ID that element.

Now, only the ID relative to URL path shows.

A very common use case for this is when clicking table of content items.

For exmple, from ourbigbook.com/barack-obama/mathematics, clicking the ToC item for "Calculus" would previously lead to ourbigbook.com/barack-obama/mathematics#barack-obama/calculus

After this change it leads just to: ourbigbook.com/barack-obama/mathematics#calculus, without repeating the "#barack-obama part as it already appears in the URL path /barack-obama/mathematics.

Short URLs were already used on Static website publishes, and weren't implemented on OurBigBook Web yet simply because this is hard. The reason this was much harder to implement on Web is that due to Dynamic article trees we can't know at render-time what the correct fragment will be, as it depends on what shows on the page or not.

And furthermore, articles by different users can appear on the same page due to topics.

The simple but not ideal solution that we were using up to now was to just have full IDs on every HTML element, make every a point to an absolute ID like /barack-obama/mathematics, and then use JS effect to hack that to #barack-obama/mathematics if the element is in the page.

What we did now is to take the Js hacks one step further, and actually replace the "long URLs" with short ones. This was not easy, partly because the browser interfaces are not amazing in that area, partly due to fighting with React. But we manage to get it working mostly well.

Announcements:

Docs: Section "Horizontal rule"

Behold:

We're experimenting with a more traditional and boring "dark" theme than the green on black classic previously used.

Readability is probably slightly better, though it is hard to measure these things. It is quite possible that the change matter much more for some people than others who have different eye sight phenotypes.

Perhaps the most important outcome of this is that it will greatly reduce the endless complaining from the community. Though perhaps that was a feature rather than a bug?

Beyond the theme change, many other changes were made. Many of those improvements feel like undisputable upgrades, e.g.:

The CSS code was also refactored and it is not much easier to make broad color changes such as these in the future, as color constants are not more closely grouped, and fewer constants are now used.

Large parts of this change were pushed forward by sidstuff who contributed a several code snippets and ideas to it.

It is now possible for admins pin an article to the homepage. The initial use case is to help with new user onboarding. Documentation: pinned article.

Previously we would only create an entry in the _file output directory for headers marked wiht the \H file argument.

For example the file file_demo/hello_world.js in this repository has an associated header with the file argument in our index.bigb :

As a result, when doing a split header conversion, it would get both:

On the other hand, the test file file_demo/nofile.js has no such associated header in the source code.

Before this change, file_demo/nofile.js would only get an _raw directory entry under _raw/file_demo/nofile.js and not _file entry. But now it also gets both.

The advantages of a _file entries over _raw entries are as follows:

This change is especially powerful following Always show large text files on _file split headers.

Because we now have _file entries for every single file, we have also modified _dir directory directory listing pages to link to _file entries as those are generally more useful than _raw which is what they previously linked to. And you can always reach _reaw_ from the corresponding _file is needed. Example: docs.ourbigbook.com/_dir

Previously, large files with an \H file argument associated to them would show a messagerather than the file contents both on their split and non-split versions, e.g.:

Now, the split version docs.ourbigbook.com/_file/index.js alwayws shows the full text file.

When not in split mode, limiting preview sizes is important otherwise multi-header pages might become far too big. Ideally we would have found a way to reliably use iframe + loading="lazy" to refer to the file without actually embedding it into the page as we do for images, but we haven't managed to do that so far.

This allows us to now see files that were previously not visible anywhere on the rendered HTML without download due to _raw files are downloaded rather than displayed in browser for certain file extensions on GitHub Pages.

The main focus was the Table of contents rendering, which had a lot of redundant stuff. Headers were the next largest gain.

The main techniques used to reduce size were:

After this changes, the rendered size of cirosantilli.com fell from 216 MiB to 156.5 MiB, which is kind of cool!

In previous updates we added insane topic links which allow you to write #mathematics to link to OurBigBook Web topics such as: ourbigbook.com/go/topic/mathematics

The outcome of that however is that it is also easy and correct to create links to topics that don't yet exist on the OurBigBook Web instance.

To make this nicer, we've unconsciously copied Wikipedia once again, and added a "Create an article for this topic" link

For example, currently OurBigBook.com the topic "Endoplasmatic Reticulum" does not have any articles on it. So if you created a link <#endoplasmatic reticulum>, it would redirect you to: ourbigbook.com/go/topic/endoplasmic-reticulum

Previously, this would show "Topic does not exist". But now it shows a button that opens the new article editor with pre-filled title "Endoplasmatic reticulum". The title choice is only a heuristic as it can't know the correct capitalization, but it covers most cases corectly by default and can be modified manually as needed.