Log for pom.xml

• Bump to tablier 0.6.2 💬 by Caio 5 years ago

  Now we have a logo that's less shameful
  

• [maven-release-plugin] prepare for next development iteration by Caio 5 years ago
• [maven-release-plugin] prepare release 0.0.1 by Caio 5 years ago
• Point at released snapshot dependencies by Caio 5 years ago
• Switch from logback to log4j2 💬 by Caio 5 years ago

  This patch configures spring boot to use log4j2 instead of logback
  for its internal logging.
  
  Additionally, loggers are configured as they were with logback but
  the console output is changed to match spring boot's default layout.
  

• Point at the next (snapshot) tablier by Caio 5 years ago
• Adjust to new diet specification format 💬 by Caio 5 years ago

  This patch upgrades cerberus, which changes the SearchQuery model
  to replace `dietThreshold()` with `diet()` (with an embedded
  threshold). With this change, it became trivial to get rid of
  the `&science` parameter oddity and shoehorn a tiny bit more of
  type checking.
  

• Add the current commit id to metric tags 💬 by Caio 5 years ago

  This patch re-enables the git-commit-id plugin execution, sets it up
  so that it writes the application.properties:git.head.describe value
  and then customizes the meter registry settings to include the new
  tag.
  
  The "application" tag introduced on bb361aa5 has been moved to be
  defined on code as well, so the common tags configuration only happens
  at one place.
  

• Upgrade to tablier 0.6.0 by Caio 5 years ago
• Upgrade to resilience4j 0.15.0 💬 by Caio 5 years ago

  It's quite annoying that the reactor support pulls vavr still,
  but hey...
  

• Bump to cerberus 0.7.1 by Caio 5 years ago
• Point at the new gula package releases by Caio 5 years ago
• Always preserve filters 💬 by Caio 5 years ago

  This patch points at tablier's interim release so that we can finally
  enable a feature where every filter option is preserved.
  
  What this means is that, now, if you search from a Search Result page,
  any selected filter will be preserved for the subsequent search. This
  is particularly helpful for a use-case I've been needing for a while:
  
  Bookmark-able, reuse-able urls. One may now bookmark a search for
  their preferences (say: `-peanut diet:keto sort:total_time`) and now
  they can use it directly as if it were a specialized site.
  

• Switch to cerberus' new SimpleRecipeMetadataDatabase by Caio 5 years ago
• Sync up with gula-bom 0.0.5 by Caio 5 years ago
• Bump to cerberus 0.6.2 by Caio 5 years ago
• Bump jsoup to 1.12.1 by Caio 5 years ago
• Bump to tablier 0.5.2 by Caio 5 years ago
• Bump spring-boot to 2.1.5-RELESE by Caio 5 years ago
• Bump tablier to 0.5.1 by Caio 5 years ago
• Bump cerberus to 0.6.1 by Caio 5 years ago
• Point at the released new cerberus and tablier by Caio 5 years ago
• Cache search queries 💬 by Caio 5 years ago

  Now that I'm more confident that we can operate _without_ a cache,
  let's add one :-)
  
  This patch adds a dependency to caffeine 2.7.0 (current) and makes
  the RecipeSearchService use a cache instance configured via the
  `cerberus.cache-size` property.
  

• Allow arbitrary queries, even empty ones 💬 by Caio 5 years ago

  This patch makes the search &q= parameter optional, so that one can
  search without typing anything (or just white-space, or "*") and
  browse all recipes. It's useful for answering questions like:
  
  > What is the quickest paleo meal to prepare
  
  This pulls in `cerberus-0.5.3-SNAPSHOT` since the 0.5.2 release
  contains logic that does not allow a query to be built without
  the `fulltext` field set.
  

• Point at tablier 0.4.6-SNAPSHOT by Caio 5 years ago
• Bump to cerberus 0.5.2 by Caio 5 years ago
• Bump cerberus to 0.5.1 💬 by Caio 5 years ago

  Smaller index size is always a win
  

• Use a fancy SearchPolicy to rewrite too-expensive queries 💬 by Caio 5 years ago

  This patch switches the policy to the new interface and makes it rewrite
  queries that are too expensive into a simple MatchAllDocsQuery(). The
  approach comes from experimentation:
  
  Terms like "tender", "lemon" and "skillet" each hit about 200k documents
  in the index currently. Searching for any combination of these ("tender
  lemon", for example), however, is slower than scanning the index
  (about 10% less throughput); So our policy now converts a query for
  such heavy terms into a simple query that matches all documents.
  
  Ideally this will: (1) almost never happen and (2) notify the user when
  it does, but for now all that we do is silently rewrite.
  
  This patch also bumps cerberus to 0.5.0, which is two feature releases
  above the previous stable one. 0.4.x released the yet-unused
  `findSimilar` (It's a WIP in tablier) searcher API.
  

• Bump spring-boot to 2.1.4 by Caio 5 years ago
• Bump resilience4j to 0.14.1 by Caio 5 years ago