Description of problem: Searching for objects that do have a specific tag in the search bar does not work. (eg. "Vms: tag = tag1") searching for objects that do not have a tag is not possible (e.g. "Vms: tag != tag1" returns nothing). Version-Release number of selected component (if applicable): 3.5.0-0.32.el6ev How reproducible: Always Steps to Reproduce: 1. Create two tags "tag1" and "tag2". 2. Assign "tag1" to vm1 and "tag2" to vm2. 3. Search for "Vms: tag != tag1" in the search bar. Actual results: Nothing is displayed. Expected results: At least vm2 is displayed. Additional info: It looks like duplicate of bug 1092609 but it really does not work even in latest released 3.5.0 version.
Created attachment 1005795 [details] screenshot with tag != filter There is at least 67 VMs matching this filter but none of them is shown.
Please clarify: Your $subject says "that do not have a tag" while your $Description relates to VMs that do have a tag
Created attachment 1005798 [details] screenshot with incorrect filter results I'm sorry, I messed up while cloning the previous bug. The problem actually is that filter 'tag != something' does not match VMs which do not have any tag. Actual result: The filter returns single VM which has a tag different than 'ci'. Expected result: The filter should return more than 67 matching VMs which do not have tag 'ci', including VMs without any tag.
This might not be a regression if bug 1092609 is considered to be distinct than this one.
I consider current behavior really unintuitive. How can user know if a tag is assigned or not? And why should user care while using filter? If you insist on current behavior please add 'existential' filter to enable filtering like (tag_assigned && tag != tag1) || no_tag_assigned. Use case for this is simple report generation: I want to generate report with VM usage per user and exclude all VMs tagged with 'CI' tag from consideration. (I do this via Python SDK.) Thank you!
(In reply to Petr Spacek from comment #7) > I consider current behavior really unintuitive. How can user know if a tag > is assigned or not? And why should user care while using filter? > > If you insist on current behavior please add 'existential' filter to enable > filtering like (tag_assigned && tag != tag1) || no_tag_assigned. > > Use case for this is simple report generation: > I want to generate report with VM usage per user and exclude all VMs tagged > with 'CI' tag from consideration. (I do this via Python SDK.) > > Thank you! There is another RFE to support NULL or empty ("") values in the search engine , this will cover the cases you mentioned and those described in the bug.
Sorry, I have to reopen the bug. The problem described in comment #3 is really present in version 3.5.0-0.32.el6ev.
Please redo the tag attachment using REST API and test if that occurs , I has opened a separate BZ for not been able to attach a tag from UI https://bugzilla.redhat.com/show_bug.cgi?id=1205697
I re-created tag via API and it made no different. Non-tagged VMs are still not returned by filter 'tag != ci'.
Okay than. Please close this bug as suplicate of the one you mentioned in commment 6 so I can track it. Thank you!
(In reply to Petr Spacek from comment #18) > Okay than. Please close this bug as suplicate of the one you mentioned in > commment 6 so I can track it. Thank you! It is not duplicate , 1205697 is a new issue not related to this one , so closing as WORKFORME
Eli, WORKSFORME does not sound appropridate for bugs which are reproducible. Please close it either as WONTFIX or as a duplicate of something, e.g. the general "NULL or empty ("") values" filter you mentioned earlier: (In reply to Eli Mesika from comment #8) > There is another RFE to support NULL or empty ("") values in the search > engine , this will cover the cases you mentioned and those described in the > bug. What is the bug # anyway? I would like to track it.
*** This bug has been marked as a duplicate of bug 1205697 ***
Eli, could you *please* point me to the RFE you mentioned in comment #8 and mentioned again in comment #21? Thank you!
Did not find it, sorry, but I remember that there was such RFE Oved, can you look in future features if the RFE for NULL/empty values support in search engine exists
(In reply to Eli Mesika from comment #24) > Did not find it, sorry, but I remember that there was such RFE > Oved, can you look in future features if the RFE for NULL/empty values > support in search engine exists Perhaps "Bug 1067844 - [RFE] add support for special values NULL and EMPTY_STRING in search mechanism" ?
(In reply to Oved Ourfali from comment #25) > Perhaps "Bug 1067844 - [RFE] add support for special values NULL and > EMPTY_STRING in search mechanism" ? Yes, this is it
Oved, thank you! *** This bug has been marked as a duplicate of bug 1067844 ***