XnView Software

Many people help behind or in the scene to make testing work properly.

So far, testing works very well. Some minor issues:

- Verification of Bug fix by original tester
I've noticed that problems fixed in Beta 2 have been verified by other people than the original tester (i.e., the person who has reported the problem). This can help, but in first instance the original tester should do the verification. This will ensure that the problem really has been fixed.

If the original tester hasn't verified fixing after 2-3 days, other people can help and
1 - reproduce the problem in beta1 and remember the exact steps and scenario.
2 - Verify the fix in beta 2 with the same steps as in previous step.

Step 1 is very important, otherwise you cannot 100% say that the bug reported has been fixed in Beta1.

- New problem in new topic
If you verify a bug fix and it's fixed, you write something like "Fixed in Beta2". Sometimes, a new problem which is a bit related is found while verifying. Please start a new topic for the new problem.

Heh, I feel somewhat responsible for the above post

If we want enforce some defect workflow, we should set up some defect management system; first link in Google http://www.laatuk.com/tools/defect_track_tools.html. It is really easier in operation and less cumbersome than forum-based.

Veryfying by user who reported- I think it would not work and it is not necessary. Unneeded waiting, ovedue checks etc. Checking in b1- I am not sure, this (veryfying in earlier version) is not a common practice. (Edit: OK, there is regression testing, but it is something a bit different anyway.) If bug happens only for some users, this should be detected in error definition stage.

Bugs should also contain detailed, distinct procedure to reproduce bug. All error reporting and verification should be run with deleted xnview.ini, except for the cases related to .ini file of course.

So, who's going to set up some DMS?

It you don't mind, I will continue to work my way, but adding "for me" phrase.

X.

Edit: Oh yes, new problems should be reported in new post, sorry for that.

Xyzzy wrote:Heh, I feel somewhat responsible for the above post

You've noticed

Xyzzy wrote:If we want enforce some defect workflow, we should set up some defect management system; first link in Google http://www.laatuk.com/tools/defect_track_tools.html. It is really easier in operation and less cumbersome than forum-based.

I fully afree. The forum based system kind of works, but is some work for moderators. Overall handling, tracking, and reporting is not so easy as in a real DTS (Defect tracking system).

Xyzzy wrote:Veryfying by user who reported- I think it would not work and it is not necessary. Unneeded waiting, ovedue checks etc. Checking in b1- I am not sure, this (veryfying in earlier version) is not a common practice. (Edit: OK, there is regression testing, but it is something a bit different anyway.) If bug happens only for some users, this should be detected in error definition stage.

Reproducing a bug in earlier version and verify fix in current version is very essential, especially if someone other than the original poster does verification.

Xyzzy wrote:Bugs should also contain detailed, distinct procedure to reproduce bug. All error reporting and verification should be run with deleted xnview.ini, except for the cases related to .ini file of course.

So, who's going to set up some DMS?

For 1.82 Testing it's too late. Perhaps for 1.83... Sören can setup any system, but we have to evaluate and find a good and suitable one.

Xyzzy wrote:It you don't mind, I will continue to work my way, but adding "for me" phrase.

O.k.

Xyzzy wrote:Edit: Oh yes, new problems should be reported in new post, sorry for that.

Someone must give an example how not to do it