<op bin="not">Text</op> doesn't exclude bugs containing 'Text'

Bug #85152 reported by Henrik Nilsen Omma
2
Affects Status Importance Assigned to Milestone
Bug Helper
Invalid
High
Unassigned

Bug Description

The attached file (when formed correctly) should exclude bugs with the word Traceback, but doesn't.

Perhaps I've not understood how the and and/or "not" works :)

Revision history for this message
Henrik Nilsen Omma (henrik) wrote :
Revision history for this message
Daniel Holbach (dholbach) wrote :

Definitely confirmed. ./bughelper -A -p ubiquity --format=html spits out bug 76660 and bug 64519 (for example) that should NOT match.

Changed in bughelper:
importance: Undecided → High
status: Unconfirmed → Confirmed
Revision history for this message
Daniel Holbach (dholbach) wrote :

In Bug 76660 it's http://librarian.launchpad.net/5499606/partman that matches, not the whole bug. (So bughelper finds a VALID match).

Revision history for this message
Daniel Holbach (dholbach) wrote :

In bug 64519 it's http://librarian.launchpad.net/4701639/partman that matches.

I'm somehow inclined to close the bug.

Can you find cases where the behaviour is really broken?

Changed in bughelper:
status: Confirmed → Needs Info
Revision history for this message
Markus Korn (thekorn) wrote :

I think I found the problem, to solve this i need to have a closer look at the code of bughelper and infoFiles.py.

The search result for each element (bug text and attachments) is not saved. Only the last element (in most cases a attachment) is responsible whether a bug is shown or not.
Example:
====================================BUG: 44595
FAT32 0
res: ( True =>show) False
Traceback 3
res: ( False =>show) True
====================================BUG: 49654
FAT32 0
res: ( True =>show) False
Traceback 1
res: ( False =>show) True
FAT32 0
res: ( True =>show) False
Traceback 2
res: ( False =>show) True
FAT32 0
res: ( True =>show) False
Traceback 1
res: ( False =>show) True
FAT32 2
res: ( True =>show) True
Traceback 0
res: ( False =>show) False
FAT32 74
res: ( True =>show) True
Traceback 0
res: ( False =>show) False
http://launchpad.net/bugs/49654 [ubiquity Ubuntu: Rejected/High] - Scan Fat32 on boot
  - http://librarian.launchpad.net/3175204/partman
  - http://librarian.launchpad.net/3175208/syslog

====================================BUG: 59257

Revision history for this message
Markus Korn (thekorn) wrote :

I found an example:
the added cluefile searches for all bughelper bugs without "suggestion."
In bug 79136 is "suggestion." in the first comment but not in the last attachment => this bug is shown

Revision history for this message
Emilio Pozuelo Monfort (pochu) wrote :

This should be a 0.2 block, IMHO

Revision history for this message
Markus Korn (thekorn) wrote :

This needs to be discussed.

Using Henrik's clue-file and looking at Bug 76660 I get the following result:

$ ./bughelper -A -p ubiquity
...
http://launchpad.net/bugs/76660 [ubiquity Ubuntu: Rejected/Undecided] - Scan Fat32 on boot
  - http://librarian.launchpad.net/5499605/syslog
  - http://librarian.launchpad.net/5499606/partman
...

syslog and partman are listed because they don't contain "Traceback"

Maybe a output like that would be better:

$ ./bughelper -A -p ubiquity
...
http://launchpad.net/bugs/76660 [ubiquity Ubuntu: Rejected/Undecided]
  - http://librarian.launchpad.net/5499605/syslog - Scan Fat32 on boot
  - http://librarian.launchpad.net/5499606/partman - Scan Fat32 on boot
...

Markus

Revision history for this message
Daniel Holbach (dholbach) wrote :

What part would you display behind the URL of the attachment that matches?

Revision history for this message
Markus Korn (thekorn) wrote :

I think we should handle the bugreport (report + comments) and the attachments as different objects. Behind each object we should display the "<info>" of each specific matched clue.

Example:

** Cluefile **
<?xml version="1.0"?>
<clues version="0.2">
  <clue>
    <contains>
      <op>foo</op>
    </contains>
    <info>foo</info>
  </clue>
  <clue>
    <contains>
      <op>bar</op>
    </contains>
    <info>bar</info>
  </clue>
</clues>

** Bugreport **
"This bug is foo"

** Attachment no. 1 **
"here is foo"

** Attachment no. 2 **
"foo and bar"

The result should be like this:
...
http://launchpad.net/bugs/xxx [ubiquity Ubuntu: Rejected/Undecided] - foo
  - http://librarian.launchpad.net/yyy/no1 - bar
  - http://librarian.launchpad.net/zzz/no2 - foo, bar
...

Please understand this as a opinion of somebody who is not so experienced with bugtriaging :)

Revision history for this message
Daniel Holbach (dholbach) wrote :

I'm not sure I understand what the problem is and how a fix should look like. Can somebody sum up the discussion somehow? Is this merely an output problem?

Revision history for this message
Launchpad Janitor (janitor) wrote :

[Expired for Bug Helper because there has been no activity for 60 days.]

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.