Hello,
Using "Text" fields for exact matching in JIRA is a bad idea. JIRA processes all the text within a "text" fields that will generally make exact matching impossible. I guess you have two options:
- Can you use the date picker custom field instead? In this field you can search for an "exact day".
- You can set your indexing language to other and do a re-index. This will mean that searching through comments, summaries and other text fields will no longer be smart. This will make searching for issues via the summary or comment field much harder.
Details
"~" actually means do a full text search with Lucene. It basically means "try and find all issues whose weekField contains text relevant to 'week-2010-11-15'".
How it does this is quite complicated and I am no expert on the topic. Basically, we first index the issue and all the text from the "text fields" (e.g comments, summary, description,...). The text is passed through a filter that removes common words and process the text in a way that makes it easy to do a full text search (e.g. it stems the text). It then stores the processed text in the index so that it can be searched later. In you case when using the Russian indexer it is actually only storing "week" in the index. The numbers at the end are dropped by the Russian indexer (not sure why).
When you do a search weekField ~ 'week-2010-11-14' JIRA (aka Lucene) runs the 'week-2010-11-14' through the same processor that was used when the issue was indexed to get 'week' essentially giving the query weekField ~ 'week'. JIRA then searches though the index for issues whose weekField has 'week'. You will not be able to find '2010-11-14' because it is neither indexed or searched.
This whole process will make searching through text fields more "Google" like, however, it makes searching for exact matches almost impossible in "text fields". Different indexing languages behave in different ways. Different versions of JIRA behave in different ways.
Regards,
Brenden.
there is a simliar issue with the german analyser: just searching for the number of an issue ID in a text field wont return the issue. u have to provide the full issue id. e.g.
text ~ 4587 wont find an issue where "MYPROJ-4578" is contained in a text field but the search for text ~ MYPOROJ-4587 does.
this is not very "google like" search and IMO should be fixed as the number part is clearly a separate token. it is just a matter on how to config the analyser chain.