Mantis Bugtracker 

View Issue Details Jump to Notes ] Issue History ] Print ]
IDProjectCategoryView StatusDate SubmittedLast Update
0007901TestLinkSecurity - Generalpublic2017-03-27 11:422017-12-01 07:36
Assigned To 
PlatformOSOS Version
Product Version1.9.14 (2015 Q3) 
Fixed in Version 
Summary0007901: using ghost steps can change executed testcases
Descriptionuse a ghoststep in step1 of TC1 with a references to a step1 of TC2.
execute TC1.
change step1 of TC2.
as you can see, step1 of TC1 will be changed to, even if TC1 has a testresult already.
if you have a testresult for a testcase, its normal that you have to create a new version of this testcase, but the ghoststeps make it possible to manipulate everything.
there are two options only.

1. eliminate ghoststeps. they can cause a lot of trouble for example recursive loops or changing the prefix of a project.

2. make sure that nobody can manipulate results with ghoststeps.

this is a real BLOCKER for TestLink
Steps To Reproducei described the steps in description
TagsNo tags attached.
Database (MySQL,Postgres,etc)MySQL
PHP Version5.4.40
QA Team - Task Workflow StatusTBD
Attached Files

- Relationships

-  Notes
fman (administrator)
2017-03-27 12:50

If you have use ghost you wants that changes on source reflects on destination.
stef-lhm (reporter)
2017-03-27 12:55

can you explain please. i am sorry, but i do not understand your comment.
do you speak german?
fman (administrator)
2017-03-27 13:36

I do not speak german.
I will retyr

you use ghost because you want to do change in a place and want this change affect multiple places, that's the idea.

Then only option that is near impossible to implement is to block SOURCE MODIFICATION (the single place) when test case that point (wants to reuse) has been executed.
BUT I do not have a simple way to get this info => can not be done

hope now is clear
stef-lhm (reporter)
2017-03-27 13:54

i understand you.

i did software developing as well and i thought the same like you thought. it is hardly impossible to implement that this source modification will be blocked, if the testcase was executed. But that means you cannot implement ghoststeps as well.
the sense of a tool like testlink is, to do a revision anytime to what have been tested. but if you can change testcases even after execution, in the way i described, you never can be sure about the testresults.
and by the way, this is the reason why executed testcases need a new version, if you like to change a testcase.
fman (administrator)
2017-03-27 19:51

You can fix this (partially) with procedures:
1) Test Cases that are SOURCE for GHOST can not be executed.
2) Test Cases that are SOURCE for GHOST can not be edited Anymore after you have changed it's status (this can be done on TL via code)

I would like to know how other Test Case management systems that implement test case reuse cope with this problem.

Problem is worst when using ghost at test step level, because you can create a test case that uses step from several test cases.

Then idea is use ghost feature best as you can
stef-lhm (reporter)
2017-03-28 12:29

in my opinion, using ghost steps needs an own table in database to show all relationships/dependencies AND to avoid:
- endless loops with each other calling ghoststeps
- problems with rename project prefix
- testcase changings in a version where testexecutions allready exist

i talked to all of my testmanagers and they see the same problems.
long time ago, we decided to use a comment after a ghoststep with the source of a step, to make sure that we see all ghoststeps.
maybe this will be a nice feature for a future testlink.
but the 3 mainproblems i mentioned are still the same.
fman (administrator)
2017-03-28 14:49

This need lot of analysis before any attempt to do a new design.
An for this an sponsorship will be a must
stef-lhm (reporter)
2017-11-30 10:01

Another problem occurred. If you use multistep ghoststeps the performance of TestLink is going down exponential.
the next problem is renumbering the steps of a case, if you use ghoststeps referencing to steps of the same testcase.

In this topic many many problems occurred while using ghosts. I designed strict rules for using ghosts for all of our users to avoid problems with TL. But another solution should be prefered.
fman (administrator)
2017-11-30 20:10

>> But another solution should be prefered.
as said in #26216 more details are needed
stef-lhm (reporter)
2017-12-01 07:36

I know, but i just wanna list the effects in just one issue. can you put a relation between these ghost step issues?

- Issue History
Date Modified Username Field Change
2017-03-27 11:42 stef-lhm New Issue
2017-03-27 12:39 fman QA Team - Task Workflow Status => TBD
2017-03-27 12:39 fman Priority immediate => high
2017-03-27 12:50 fman Note Added: 0026204
2017-03-27 12:55 stef-lhm Note Added: 0026205
2017-03-27 13:36 fman Note Added: 0026206
2017-03-27 13:54 stef-lhm Note Added: 0026209
2017-03-27 19:51 fman Note Added: 0026212
2017-03-28 12:29 stef-lhm Note Added: 0026215
2017-03-28 14:49 fman Note Added: 0026216
2017-11-30 10:01 stef-lhm Note Added: 0027038
2017-11-30 20:10 fman Note Added: 0027042
2017-12-01 07:36 stef-lhm Note Added: 0027045

Copyright © 2000 - 2020 MantisBT Team
Powered by Mantis Bugtracker