A public update to all watchers/voters, the remaining work for this fix to be complete is complete and the fix should be delivered in JIRA 6.2.3 and the next JIRA OnDemand update.

Regards,

Oswaldo Hernández.
JIRA Bugmaster.
[Atlassian].

Happy 8 Year Anniversary to the oldest in-progress ticket here!

I think I speak for many of us when I say we are still anxiously awaiting a fix to this issue.

Hi all,

Just an update on this issue, we recently started work on the problems outlined in this issue. After investigating this, it was clear that JIRA's text analysis chain was not designed with these exact text phrase queries in mind.

In consequence, we have had to divide the work in three issues which progressively improve the current situation:

• Apply a different analysis algorithm at query time that does not stem the query text when the exact phrase text search syntax is used (JRA-34211)
• Alter our field indexing strategy so that we index an additional Lucene field for exact phrase searches which contains no stemmed content (same analysis algorithm used at query time), and have these types of search use that field instead of the standard field. (JRA-34212)
• Alter the exact phrase analysis chain so that stop-word removal is not done both at query/index time for these type of searches. (JRA-34218)

In JIRA 6.0.7 we will be delivering the improvement outlined in JRA-34211, and currently I am starting work on JRA-34212.

The only workaround we have for this at the moment is to disable stemming by setting your indexing language to none. We'll revisit this issue when we look at the JQL theme.

Is there an option to do this optional? The user should have both options for search. Like:

Find all issues where the Summary contains the exact phrase "full screen"

summary ~ "\"full screen\""

But even here we get also issues which have only part of this words.

Thank you for the reply and considerations.

Hi sharon.lindeblad@gtech.com,

I'm glad to hear that. Personally, I haven't conducted sufficient testing to be 100% certain about other possible cautions...

What I can say is that after switching JIRA's Indexing language option to Other (+ re-indexing JIRA), if any of your existing filters intentionally:

• search for 'stemmable' words in any field (i.e. which will retrieve more issues than those expected from exact matches), then the results retrieved by those filters may be reduced (since searches will now rely on exact matches).
• search for reserved words in text fields, then the results retrieved by those filters may increase (since reserved words will now be indexed).

Cheers,
Giles.

Giles,

Thanks very much for the suggestion.

I just tried that on my test server and it worked perfectly for exact phrase match search.
Are there any cautions for this? i.e. will it adversely affect existing filters?

Thanks much,
Sharon

Have you tried switching JIRA's Indexing language option to Other and then re-indexing your JIRA site? The documentation for this has just been updated accordingly.

This should effectively switch JIRA to an "exact phrase" mode.

Pretty basic functionality to have in a tracking tool…
I’m adding my voice; fix this!
Please.