#131: Correct parentheses support in in search parsing
------------------------+---------------------------------------------------
  Reporter:  jblayloc   |       Owner:  jblayloc                       
      Type:  defect     |      Status:  assigned                       
  Priority:  critical   |   Milestone:                                 
 Component:  WebSearch  |     Version:                                 
Resolution:             |    Keywords:  INSPIRE Invenio Syntax News Oct
------------------------+---------------------------------------------------

Comment (by simko):

 Replying to [comment:13 tbrooks]:
 > So I just went through and confirmed that 2 a,b,c are all behaving the
 same on a machine running Joe's branch, inspirebeta, and invenio-
 demo.cern.ch

 Actually there is a difference in behaviour: e.g. query `(ellis (NOT
 muon))` with master branch reports that parens are nested, so going to be
 dropped, and the query returns 11 hits.  With Joe's branch, no message is
 emitted, but no results are returned either, one gets basically an empty
 page.  This is less user-friendly than what the master branch does.

 > I'd propose to get that moving along, and create new tickets for these
 additional fixes.

 If these are easy to fix, then let's fix them now as part of the
 "//correct// parens support" ticket.  If they are less easy, then let's
 fix at least the empty results behaviour.  If this require deeper changes,
 then let's consider even this separately indeed -- but we should at least
 emit something like "Sorry, no hits found, please avoid using nested
 parentheses in the query" or the like in this branch. (Of course, only
 when no hits are printed and when query used nested parens.)

 > When creating new tickets it is also worth understanding why the
 regression tests for the (U(1) OR SL(2,Z)) expects a different result with
 or without parens.

 This is indeed a bug that was mistakenly introduced by
 7b3804d29a49dd86a64eedba9714795244286653.  I have just fixed it. Thanks.

-- 
Ticket URL: <http://invenio-software.org/ticket/131#comment:16>
Invenio <http://invenio-software.org>

Reply via email to