This page contains changes which are not marked for translation.
This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Improvements and Maintenance
Editors interested in templates can help by reading the latest Wishlist focus area, Template recall and discovery, and share your feedback on the talkpage. This input helps the Community Tech team to decide the right technical approach to build. Everyone is also encouraged to continue adding new wishes.
The new automated Special:NamespaceInfo page helps editors understand which namespaces exist on each wiki, and some details about how they are configured. Thanks to DannyS712 for these improvements. [1]
References Check is a feature that encourages editors to add a citation when they add a new paragraph to a Wikipedia article. For a short time, the corresponding tag "Edit Check (references) activated" was erroneously being applied to some edits outside of the main namespace. This has been fixed. [2]
It is now possible for a wiki community to change the order in which a page’s categories are displayed on their wiki. By default, categories are displayed in the order they appear in the wikitext. Now, wikis with a consensus to do so can request a configuration change to display them in alphabetical order. [3]
Tool authors can now access ToolsDB's public databases from both Quarry and Superset. Those databases have always been accessible to every Toolforge user, but they are now more broadly accessible, as Quarry can be accessed by anyone with a Wikimedia account. In addition, Quarry's internal database can now be queried from Quarry itself. This database contains information about all queries that are being run and starred by users in Quarry. This information was already public through the web interface, but you can now query it using SQL. You can read more about that, and 20 other community-submitted tasks that were resolved last week.
Any pages or tools that still use the very old CSS classes mw-message-box need to be updated. These old classes will be removed next week or soon afterwards. Editors can use a global-search to determine what needs to be changed. It is possible to use the newer cdx-message group of classes as a replacement (see the relevant Codex documentation, and an example update), but using locally defined onwiki classes would be best. [4]
Technical project updates
Next week, all Wikimedia wikis will be read-only for a few minutes. This will start on September 25 at 15:00 UTC. This is a planned datacenter switchover for maintenance purposes. This maintenance process also targets other services. The previous switchover took 3 minutes, and the Site Reliability Engineering teams use many tools to make sure that this essential maintenance work happens as quickly as possible. [5]
Tech in depth
The latest monthly MediaWiki Product Insights newsletter is available. This edition includes details about: research about hook handlers to help simplify development, research about performance improvements, work to improve the REST API for end-users, and more.
To learn more about the technology behind the Wikimedia projects, you can now watch sessions from the technology track at Wikimania 2024 on Commons. This week, check out:
Hackathon Showcase (45 mins) - 19 short presentations by some of the Hackathon participants, describing some of the projects they worked on, such as automated testing of maintenance scripts, a video-cutting command line tool, and interface improvements for various tools. There are more details and links available in the Phabricator task.
Co-Creating a Sustainable Future for the Toolforge Ecosystem (40 mins) - a roundtable discussion for tool-maintainers, users, and supporters of Toolforge about how to make the platform sustainable and how to evaluate the tools available there.
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Weekly highlight
All wikis will be read-only for a few minutes on Wednesday September 25 at 15:00 UTC. Reading the wikis will not be interrupted, but editing will be paused. These twice-yearly processes allow WMF's site reliability engineering teams to remain prepared to keep the wikis functioning even in the event of a major interruption to one of our data centers.
Updates for editors
A screenshot of the interface for the Alt Text suggested-edit feature
Editors who use the iOS Wikipedia app in Spanish, Portuguese, French, or Chinese, may see the Alt Text suggested-edit experiment after editing an article, or completing a suggested edit using "Add an image". Alt-text helps people with visual impairments to read Wikipedia articles. The team aims to learn if adding alt-text to images is a task that editors can be successful with. Please share any feedback on the discussion page.
The Codex color palette has been updated with new and revised colors for the MediaWiki user interfaces. The most noticeable changes for editors include updates for: dark mode colors for Links and for quiet Buttons (progressive and destructive), visited Link colors for both light and dark modes, and background colors for system-messages in both light and dark modes.
It is now possible to include clickable wikilinks and external links inside code blocks. This includes links that are used within <syntaxhighlight> tags and on code pages (JavaScript, CSS, Scribunto and Sanitized CSS). Uses of template syntax {{…}} are also linked to the template page. Thanks to SD0001 for these improvements. [6]
Two bugs were fixed in the GlobalVanishRequest system by improving the logging and by removing an incorrect placeholder message. [7][8]
The API now enables 5,000 on-demand API requests per month and twice-monthly HTML snapshots freely (gratis and libre). More information on the updates and also improvements to the software development kits (SDK) are explained on the project's blog post. While Wikimedia Enterprise APIs are designed for high-volume commercial reusers, this change enables many more community use-cases to be built on the service too.
The Snapshot API (html dumps) have added beta Structured Contents endpoints (blog post on that) as well as released two beta datasets (English and French Wikipedia) from that endpoint to Hugging Face for public use and feedback (blog post on that). These pre-parsed data sets enable new options for researchers, developers, and data scientists to use and study the content.
In depth
The Wikidata Query Service (WDQS) is used to get answers to questions using the Wikidata data set. As Wikidata grows, we had to make a major architectural change so that WDQS could remain performant. As part of the WDQS Graph Split project, we have new SPARQL endpoints available for serving the "scholarly" and "main" subgraphs of Wikidata. The query.wikidata.org endpoint will continue to serve the full Wikidata graph until March 2025. After this date, it will only serve the main graph. For more information, please see the announcement on Wikidata.
Here's your quick overview of what has been happening around Wikidata in the week leading up to 2024-09-30. Please help Translate. Missed the previous one? See issue #646
Wikidata's 12th birthday is coming up on October 29th. Have a look at the birthday parties and more planned around the world.
Next Linked Data for Libraries LD4 Wikidata Affinity Group session 1 October, 2024: We have our next LD4 Wikidata Affinity Group Session on Tuesday, 1 October, 2024 at 9am PT / 12pm ET / 17:00 UTC / 6pm CET (Time zone converter). Christa Strickler will be our first Project Series lead with her joint project with the Wikidata Religion & Theology Community of Practice to contribute biographical data to Wikidata from the IRFA database using the Mix’n’Match tool. We are excited to learn more about this project, provide a forum for discussion and shared learning, and lend a hand while building new skills. Event page.
Press, articles, blog posts, videos
Papers
A Systematic Review of Wikidata in GLAM Institutions: a Labs Approach - Presents a systematic review of Wikidata use in GLAM institutions within the context of the work of the International GLAM Labs Community (glamlabs.io). The results summarise academic literature on Wikidata projects. By G. Candela et al.
Using Wikidata for Managing Cultural Heritage Information - The present study uses model wikidata elements as a basis and explores its dynamic formation into a cultural heritage information management tool within a museum. By D. Kyriaki-Manessi and S. Vazaiou.
Three new Userscripts for Wikidata - User:Lagewi has written 3 scripts to simplify reading references, explore property-value pairs in use for a statement or attaching a full bibliography to the end of the item page.
objects of occurrence have role (role that objects of this occurrence take on in the context of this occurrence. (For selectional restrictions, use "object class of occurrence" (P12913) instead.))
agents of action have role (role that agents of this action take on in the context of this action. (For selectional restrictions, use "agent class of action" (P12994) instead. ))
agent class of action (class of items that may initiate this action or class of actions (For roles filled by agents of an action, use "agents of action have role" (P12993) instead))
agent of action (particular item that initiates this action or class of actions)
Larval host plant (Larval host plant - used only for insects - subclass of P1034)
has reading (phonetic reading or pronunciation of the kanji)
chemical formula (Description of chemical compound giving element symbols and counts)
mode of reproduction (ways for living organisms to propagate or produce their offsprings)
health points (health or armor points of this video game, board game or role-playing game character)
damage (damage value of this video game weapon, ability or character)
magazine capacity (In (real or fictional) devices like a firearm, weapon, or engineered thing, this is the default capacity or size of a devices' magazine, clip, or other container typically used to hold ammunition, bolts, cartridges, tools, etc. which pushes those items as needed usually through a spring-based mechanism into a receiver for further use by the device)
male mean age (male mean age in a given place; qualifier of {{P|4442}})
female mean age (female mean age in a given place; qualifier of {{P|4442}})
Western Australian Biographical Index (Card ID from the Western Australian Biographical Index, a set of handwritten index cards compiled in the 1970s.)
leased to (person or organisation that holds or was granted a lease on the subject)
WPBSA com player ID (Identifier for an athlete on the main website of WPBSA)
JLPT level (difficulty of word by the level of JLPT)
Search: The haswbstatement search magic word has been improved by the Search Platform Team. Previously it was limited in which Properties were indexed for it. Going forward haswbstatement:P123 will work for all Properties, regardless of their datatype. This will allow you to filter search results for Items that have a statement with a specific Property. (Searching for a specific complete statement with haswbstatement:P123=xxx will still only work for specific datatypes.) For this to work all Items have to be reindexed and this will take up to 1 month.
Design system migration: We have migrated the Special:NewLexeme page from Wikit to Codex and are working on finishing the migration for the Query Builder.
EntitySchemas: We finished the investigation about how to support search for EntitySchemas by label or alias when linking to an EntitySchema in a statement. (phab:T362005)
Wikibase REST API: We worked on integrating language fallbacks into the API (phab:T371605)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Updates for editors
Readers of 42 more wikis can now use Dark Mode. If the option is not yet available for logged-out users of your wiki, this is likely because many templates do not yet display well in Dark Mode. Please use the night-mode-checker tool if you are interested in helping to reduce the number of issues. The recommendations page provides guidance on this. Dark Mode is enabled on additional wikis once per month.
Editors using the 2010 wikitext editor as their default can access features from the 2017 wikitext editor by adding ?veaction=editsource to the URL. If you would like to enable the 2017 wikitext editor as your default, it can be set in your preferences. [9]
For logged-out readers using the Vector 2022 skin, the "donate" link has been moved from a collapsible menu next to the content area into a more prominent top menu, next to "Create an account". This restores the link to the level of prominence it had in the Vector 2010 skin. Learn more about the changes related to donor experiences. [10]
The CampaignEvents extension provides tools for organizers to more easily manage events, communicate with participants, and promote their events on the wikis. The extension has been enabled on Arabic Wikipedia, Igbo Wikipedia, Swahili Wikipedia, and Meta-Wiki. Chinese Wikipedia has decided to enable the extension, and discussions on the extension are in progress on Spanish Wikipedia and on Wikidata. To learn how to enable the extension on your wiki, you can visit the CampaignEvents page on Meta-Wiki.
Developers with an account on Wikitech-wiki should check if any action is required for their accounts. The wiki is being changed to use the single-user-login (SUL) system, and other configuration changes. This change will help reduce the overall complexity for the weekly software updates across all our wikis.
In depth
The server switch was completed successfully last week with a read-only time of only 2 minutes 46 seconds. This periodic process makes sure that engineers can switch data centers and keep all of the wikis available for readers, even if there are major technical issues. It also gives engineers a chance to do maintenance and upgrades on systems that normally run 24 hours a day, and often helps to reveal weaknesses in the infrastructure. The process involves dozens of software services and hundreds of hardware servers, and requires multiple teams working together. Work over the past few years has reduced the time from 17 minutes down to 2–3 minutes. [11]
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Weekly highlight
Communities can now request installation of Automoderator on their wiki. Automoderator is an automated anti-vandalism tool that reverts bad edits based on scores from the new "Revert Risk" machine learning model. You can read details about the necessary steps for installation and configuration. [12]
Updates for editors
Translators in wikis where the mobile experience of Content Translation is available, can now customize their articles suggestion list from 41 filtering options when using the tool. This topic-based article suggestion feature makes it easy for translators to self-discover relevant articles based on their area of interest and translate them. You can try it with your mobile device. [13]
It is now possible for <syntaxhighlight> code blocks to offer readers a "Copy" button if the copy=1 attribute is set on the tag. Thanks to SD0001 for these improvements. [14]
Customized copyright footer messages on all wikis will be updated. The new versions will use wikitext markup instead of requiring editing raw HTML. [15]
Later this month, temporary accounts will be rolled out on several pilot wikis. The final list of the wikis will be published in the second half of the month. If you maintain any tools, bots, or gadgets on these 11 wikis, and your software is using data about IP addresses or is available for logged-out users, please check if it needs to be updated to work with temporary accounts. Guidance on how to update the code is available.
Rate limiting has been enabled for the code review tools Gerrit and GitLab to address ongoing issues caused by malicious traffic and scraping. Clients that open too many concurrent connections will be restricted for a few minutes. This rate limiting is managed through nftables firewall rules. For more details, see Wikitech's pages on Firewall, GitLab limits and Gerrit operations.
The above templates (SD, speedy, Badname) expand to full speedy deletion requests.
In particular, this is better practice than just blanking the category page. Blanked categories end up on Special:UncategorizedCategories, which is largely handled by non-admins who cannot delete the categories in question.
If you believe the content does not meet the criteria for speedy deletion, you may replace the speedy deletion tag with a regular deletion request (if the content has not been deleted) or request undeletion (if the content has already been deleted).
All your uploads, including deleted ones, are listed in your upload log.
If you need help, please read our frequently asked questions or visit the help desk. Please do not remove this message from your talk page. You may set up archiving instead. Warning: Wikimedia Commons takes copyright violations very seriously and persistent violators will be blocked from editing.
Next Linked Data for Libraries LD4 Wikidata Affinity Group session 15 October, 2024: We have our next LD4 Wikidata Affinity Group Session on Tuesday, 15 October, 2024 at 9am PT / 12pm ET / 16:00 UTC / 6pm CEST (Time zone converter). https://zonestamp.toolforge.org/1729008000 Christa Strickler will be our first Project Series lead with her joint project with the Wikidata Religion & Theology Community of Practice to contribute biographical data to Wikidata from the IRFA database https://irfa.paris/en/en-learn-about-a-missionary/ using the Mix’n’Match tool. We are excited to learn more about this project, provide a forum for discussion and shared learning, and lend a hand while building new skills. Event page: [21]
The next Wikidata+Wikibase office hours will take place on Wednesday, 16th October 2024 at 18:00 CEST in the Wikidata Telegram group. The Wikidata and Wikibase office hours are online events where the development team presents what they have been working on over the past quarter, and the community is welcome to ask questions and discuss important issues related to the development of Wikidata and Wikibase.
Wikidata:Twelfth Birthday: We already have 30 events scheduled on the list 😍. As a reminder, when your event is ready, don't forget to:
create a wikipage with more information about the event, participants list, etc.
Small data, slow data − a SNAIL approach to Wikidata: discusses the value of small, carefully curated datasets in the era of big data. It emphasizes the importance of taking a methodical, "snail-paced" approach to data collection and analysis, which can lead to more meaningful and accurate insights. The blogpost also highlights how this approach can complement the broader trends of big data, ensuring that detailed, high-quality data is not overlooked.
Papers
"WoolNet: Finding and Visualising Paths in Knowledge Graphs" given two or more entities requested by a user, the system finds and visualises paths that connect these entities, forming a topical subgraph of Wikidata (Torres Gutiérrez and Hogan)
Dynamic Mapping using Collaborative Knowledge Graphs: Real-Time SKOS Mapping from Wikidata: This presentation introduces a workflow using SPARQL queries to dynamically map live Wikidata data to SKOS concepts, featuring a Python tool that converts CSV outputs into RDF triples for integration into linked data environments and knowledge graphs, emphasizing real-time data retrieval and interoperability.
Could making Wikidata 'human' readable lead to better AI?: Lydia Pintscher (WMDE), Portfolio Lead Product Manager at Wikidata Deutschland, discussed a new project aimed at making Wikidata more 'human' readable for Large Language Models (LLMs), which could improve AI reliability by giving these models access to high-quality, human-curated data from Wikidata.
Elemwala (এলেমওয়ালা) (https://elemwala.toolforge.org): is a proof-of-concept interface that allows you to input abstract content and get natural language text in a given output language. There may well be errors with particular inputs, and the text may not be quite as natural as you might expect, but that's where your improvements to your language's lexemes, other Wikidata items, and the tool's sourcecode come in!
mlscores: Tool for calculating multilinguality score of Wikidata items (including properties). E.g. for Wikidata (Q2013), the scores are - en: 99.66%, fr: 89.49%, es: 84.07%, pt: 68.47%. For instance of (P31), the scores are - en: 99.86%, fr: 87.12%, es: 80.83%, pt: 61.37%.
Other Noteworthy Stuff
Launch of WikiProject Deprecate P642: The goal of this project is to prepare for deprecation, and eventual removal, of the property of (P642). Currently, of (P642) is labeled as "being deprecated", meaning its use is still allowed, but discouraged. From a peak of around 900,000 uses, the property now has around 700,000 uses (see status here). Our goal is to reduce that as much as possible in a systematic way, while ensuring that appropriate properties exist to replace all valid uses of of (P642). The latter is key to officially deprecating the property. Before removing the property, we want to get as close to zero uses as possible.
bais (Indicates a specific form of bias present in a media source, organization, or document, such as false balance, slant, or omission, affecting the representation of information.)
TDK lexeme ID (Dictionary created by the [[Q1569712|Turkish Language Association]])
Atatürk Ansiklopedisi ID (Online Turkish encyclopedia created by [[Q6062914]] and [[Q19610584]])
Eurotopics ID (A database containing data on European media.)
Stated in unreliable source (used in the references field to refer to the database that is considered a unreliable source in which the claim is made)
PatternsKilkenny - Patterns were devotional days on the day of the patron saint of a parish or area or at least an annually occurring day when the people of the locality held their personal devotions in a certain pattern (hence the name), i.e. "doing the rounds" around trees or other landmarks at the sacred site. This project tries to collate the records and memories of these patterns for County Kilkenny.
Deprecate P642 - The goal of this project is to prepare for deprecation, and eventual removal, of the property of (P642).
AIDS Walks - This project aims to collaborate with Wiki editors across the globe to highlight AIDS Walks anywhere in the world.
Temples in Roman Britain - The aim of the Wikiproject Temples in Roman Britain is to record and catalog sacred spaces in the Roman province Britannia between 43 to 409 CE. By sacred spaces, we include (for the moment) only built structures such as temples, sanctuaries and shrines.
Nihongo - The goal of this project is to capture the Japanese Language Japanese (Q5287) in its entirety on Wikidata. We aim to give advice and establish standards for representing Japanese words as lexemes.
EntitySchemas: We are continuing the work on making it possible to find an EntitySchema by its label or aliases when linking to an EntitySchema in a statement (phab:T375641)
Design system: We are continuing the work on migrating the Query Builder from Wikit to Codex
REST API: We finished the work on language fallback support in the REST API (phab:T371605)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.
Updates for editors
The Structured Discussion extension (also known as Flow) is starting to be removed. This extension is unmaintained and causes issues. It will be replaced by DiscussionTools, which is used on any regular talk page. A first set of wikis are being contacted. These wikis are invited to stop using Flow, and to move all Flow boards to sub-pages, as archives. At these wikis, a script will move all Flow pages that aren't a sub-page to a sub-page automatically, starting on 22 October 2024. On 28 October 2024, all Flow boards at these wikis will be set in read-only mode. [22][23]
WMF's Search Platform team is working on making it easier for readers to perform text searches in their language. A change last week on over 30 languages makes it easier to find words with accents and other diacritics. This applies to both full-text search and to types of advanced search such as the hastemplate and incategory keywords. More technical details (including a few other minor search upgrades) are available. [24]
View all 20 community-submitted tasks that were resolved last week. For example, EditCheck was installed at Russian Wikipedia, and fixes were made for some missing user interface styles.
Updates for technical contributors
Editors who use the Toolforge tool Earwig's Copyright Violation Detector will now be required to log in with their Wikimedia account before running checks using the "search engine" option. This change is needed to help prevent external bots from misusing the system. Thanks to Chlod for these improvements. [25]
Some HTML elements in the interface are now wrapped with a <bdi> element, to make our HTML output more aligned with Web standards. More changes like this will be coming in future weeks. This change might break some tools that rely on the previous HTML structure of the interface. Note that relying on the HTML structure of the interface is not recommended and might break at any time. [27]
In depth
The latest monthly MediaWiki Product Insights newsletter is available. This edition includes: updates on Wikimedia's authentication system, research to simplify feature development in the MediaWiki platform, updates on Parser Unification and MathML rollout, and more.
The latest quarterly Technical Community Newsletter is now available. This edition include: research about improving topic suggestions related to countries, improvements to PHPUnit tests, and more.
Editors interested in templates can help by reading the latest Wishlist focus area, Template recall and discovery, and share your feedback on the talkpage. This input helps the Community Tech team to decide the right technical approach to build. Everyone is also encouraged to continue adding new wishes.
Tool authors can now access ToolsDB's public databases from both Quarry and Superset. Those databases have always been accessible to every Toolforge user, but they are now more broadly accessible, as Quarry can be accessed by anyone with a Wikimedia account. In addition, Quarry's internal database can now be queried from Quarry itself. This database contains information about all queries that are being run and starred by users in Quarry. This information was already public through the web interface, but you can now query it using SQL. You can read more about that, and 20 other community-submitted tasks that were resolved last week.
