Jump to content


This is a ready-only archive of the InstallSite Forum. You cannot post any new content here. / Dies ist ein Archiv des InstallSite Forums. Hier können keine neuen Beiträge veröffentlicht werden.
Photo

What kind of Patches do you give your QAs


3 replies to this topic

InstallDev

InstallDev
  • Full Members
  • 43 posts

Posted 18 February 2005 - 17:54

Hi all,

First I am sorry for the long post. But I thought this would also be a good post to help others learning about a possible way of doing patches.

I am wondering how people test their patches? Let’s explain what we want to do…

After releasing the Product. We would like to deliver the updates through patches to the end users. We will still build full msi packages that contain the new fixes after release. Then we will use the Patch design of installshield to create the patch between the released Product and the new full maintenance product. This will generate a Patch with a GUID. Now we want to test this patch, we send it to QA. Later but before release the patch to end user, developers add more bug fixes, and a new PATCH with a new GUID is created between the released product and the latest msi package. This patch has to have the first patch GUID in the “list of previous GUID to replace”. This will allow the QA to use this new patch without uninstalling the whole product. (This technique however will require the source of the released product msi to be present when patching, which we don’t have a problem with).

Now, I know this way of patching is correct when we send it to end-users. Because the number of patches that are send to end-users are small. However, I don’t think it is correct when we send it to QAs. Because the number of patches that are send to QA is relatively big. With this way we can end up with a patch, which has a list of hundreds of GUIDs(Assuming that we will send only patches for end users for a couple of years). Remember that these patches, which we send to QA, are the candidates to be end-users patches. I mean, when we decide to send a patch to end-users, it would probably be the last patch send to QA (the list of all GUIDs will be present in it).

My Question is, what is the best strategy to create patches that can also work for QA? How QA should test our Patches? Should we always ask QA to Unistall the product, install the Realesed product and then apply the patch…. (QAs will not like it at all)
Also, We don’t want to give QA access to the FULL maintenance msi package. Because we want them always to test the patching process the end-user would experience.

Any suggestions, comments are really appreciated.

Thank you,


Stefan Krueger

Stefan Krueger

    InstallSite.org

  • Administrators
  • 13,269 posts

Posted 21 February 2005 - 11:34

So not all patches that you send to QA will be released to end users? As a result your QA team will not test the same scenario that your end users will have (the QA team will have intermediate patches installed). I doubt that this is a good test plan.
Why doesn't your QA team use virtual machine software (VirtualPC or VMware) to reset their test machine to a defined state before testing your new patch?

InstallDev

InstallDev
  • Full Members
  • 43 posts

Posted 21 February 2005 - 17:58

Hi Stefan,

Yes, you are right... we are not actually testing the real experience the end user would be experiencing when they update the product. As a minimum, we should ask our QA team before releasing the patch to end users to do their testing by going through the sequence "uninstall Product -install released product - patch it" to make sure of correct implementation.

But till now I cannot see how others is doing this. Do you think they are asking their QA team to uninstall-install-apply each time they produce a Patch? Do you think they give them access to the FULL msi package? But then they end up with the same problem. QA is not really testing the actual experience the end user is having. Also, I don't think they are using virtual machine software . Even thought the idea looks nice, we are not sure how reliable these software are. A detailed evaluation should be done for this. Is their any other way that I missed?

Stefan, from your experience, what do you think the best way to create patches and test them???

Thank you very much.....

Stefan Krueger

Stefan Krueger

    InstallSite.org

  • Administrators
  • 13,269 posts

Posted 22 February 2005 - 14:33

From my experience, using virtual machine software is the industry standard method, and is reasonably reliable. If you want to be more sure, you would have to use a disk imaging software to reset a real machine to a known state.
The method you described in your first post is probably one of the worst possible. Running uninstall is better, but is still less reliable than virtual machines.