Discovery/Status updates/2018-10-22

This is the weekly update for the week starting 2018-10-22

Discussions

edit
edit
  • Mathew and Gehel worked on refactoring the WDQS updater while also increasing the restart interval of the WDQS updater
  • There were some known issues that the Search team helped in fixing where the Elasticsearch puppet config changes broke puppet in various instances
  • Erik fixed an issue where the sister search / cross language search interaction with multi-datacenter was a bit off
  • The team worked on getting Debian Stretch off of Ubuntu Trusty for all instances of the search project, as well as various other updates needed to get ElasticSearch ready for version 6: T206468, T206469, T206472, T206470, as well as investigating just how many number of servers we really need in the ElasticSearch cluster.
  • Gehel and Mathew also worked on modifying scap::target to define sudo rules for multiple services
  • Erik fixed a failure where cirrusSearchIncomingLinkCount was getting a 'read timeout is reached' issue from title::getFirstRevision
  • Erik finished up work to add autocomplete evaluation via MRR to relforge

Other Noteworthy Stuff

edit
  • Oct 18-21 Trey went to WikiConferenceNA 2018. [1] Some highlights:
    • He was interviewed [2] for the Wikijabber podcast. [3]
    • He gave a lightning talk on homoglyphs. Slides are available online. [4]
    • You can see the full conference schedule online [5] and find the slides for many of the presentations on Commons. [6]
    • More detailed write up on MediaWiki. [7]

Did you know?

edit
  • Homoglyphs [8] are characters in different writing systems that look very similar or, in some fonts, even identical to each other. For example A/Α/А or M/Μ/М in Latin, Greek, and Cyrillic, respectively. Homoglyphs can be very hard to detect when reading, but they interfere with search because “Мoscow” and “Moscow” are completely different words from the point of view of search. The former starts with a Cyrillic M (as does Moscow’s name in Russian, “Москва”). There are often a few dozen examples of the mixed-script “Мoscow” [9] on English Wikipedia at any given time. (Trey cleans them up from time to time as a hobby, and hopes to create a plugin for Elasticsearch to automatically make them visible to search.)

--