Jump to content

Developer Satisfaction Survey/2026/MediaWiki Core backend

From mediawiki.org

πŸ“– Developer Satisfaction 2026 Report

🌻 MediaWiki Core backend

[edit]

tl;dr

  • ↑ Component satisfaction
    • All 21 backend components were rated more positively than negatively by respondents, compared to 17 out of 21 last year.
    • >75% are satisfied with internationalization and localization, Installer/Updater, Rdms library, logging, and HTMLform.
    • Compared to last year, respondents reported more satisfaction with
      • Installer/Updater (79% satisfied, up from 50%)
      • HTMLform (71%, up from 46%)
      • User management (54%, up from 35%), and
      • ParserCache (56%, up from 42%).
  • Component improvement - >60% rated Rdms library and Parsoid (new parser) as being better than they were 1 year ago.


# Backend component satisfaction

"How satisfied are you with the following MediaWiki core backend components?"

All 21 of the backend components were rated more positively (i.e., very or somewhat satisfied) than negatively (i.e., very or somewhat dissatisfied) by respondents.

The top 10 backend components with the highest ratings (i.e., satisfied minus dissatisfied) were:

  • Internationalization and localization
  • Installer/Updater
  • Rdms library
  • Logging
  • HTMLform
  • Revision storage
  • ParserCache
  • User management (preferences, block, rename, etc.)
  • Diff

Compared to last year, which components moved up into the top 10?

  • HTMLform (satisfaction rose from 46% last year β†’ 71%, and dissatisfaction dropped from 32% β†’ 9%)
  • Installer/Updater (satisfaction 50% β†’ 79%, and dissatisfaction 19% β†’ 8%)
  • User management (satisfaction 35% β†’ 54%, and dissatisfaction 29% β†’ 9%)
  • ParserCache (satisfaction 42% β†’ 56%, and dissatisfaction 19% β†’ 4%)



top


# Backend component comparison

For the backend components that survey takers rated in the first question, we asked:

β€œHow would you describe the components now, as compared to 1 year ago?”

For most of the backend components, the majority of respondents rated them as being neither better nor worse than 1 year ago.

The backend components with the highest improvement ratings are as follows:

  • The majority (73%) of respondents who rated Rdms library said Rdms library is better than 1 year ago.
  • The majority (60%) of respondents who rated Parsoid (new parser) said Parsoid is better than 1 year ago.
  • And 47% of respondents who rated patrolling, recentchanges, and watchlist backend said these are is better than 1 year ago.

top


# Feedback

β€œPlease share any other feedback you may have about MediaWiki core backend components.”

There were 8 responses to this question. A few themes emerged:

MediaWiki ownership

IMO it is a big problem that a large number of MediaWiki core components are without a code steward. E.g. -- What happens if there's a security bug in that component? What happens if someone writes a patch for that component (ie., will there be someone to review it?)? My dream would be for more(/all) currently-unassigned components/extensions/skins to have a code steward assigned to them.

Parser-related feedback

Managing two parsers really slows us down as we often have to build products/experiments twice. The sooner we can get rid of legacy parser the better.

Code complexity

I will say that librdbms and bagostuff/cache libraries are both pretty complicated and I would be very wary of touching them at all.

Authentication is to the point where it's hard to reason about side effects of any changes

Additionally, we did get some thanks :)

Thank you for maintaining the core backend components!