Hey, so in our environment we've thinapped Mozilla's Firefox and Thunderbird (each their own capture). Capture machine is a Windows 7 SP1 32bit virtual machine with ThinApp 4.6.2 (we're only deploying to Windows 7 machines).
Today I was going to update both of these and had the strangest thing happen to me...
I think around v6 the "in-app" update stopped working and for both v6 and v7 I had to enable cmd.exe, launch that and start the installers from within the sandbox through the virtual "cmd.exe". For v7 this got complicated one step further as the Firefox installer couldn't find Firefox, and I had to extract the setup files from within the installer and then it worked.
After having applied the update to the sandbox, I usually went ahead and did a "sbmerge print" and then a "sbmerge apply" if stuff looked okay...
Not so today.
I was trying the v8 installers, and as with the previous two versions the "in-app" update function couldn't complete/run. So I decided to once again use the virtual "cmd.exe" approach. I enabled cmd.exe, launched it and was inside the sandbox... I verified this by doing a "dir c:\program files" and found the Mozilla Firefox folder there within the virtual environment... so I ran the installer, and it finished without a problem.
Now, previously when I've done this and then tried to launch Firefox the new version would start, I could configure it and then do the sbmerge thing. This time however... The old version of Firefox started! v7... not v8! A bit confused I exited and tried to start it again... same result.
At this point I started looking at the sandbox... and whoa, there was no sign of any new programfiles or installations in the sandbox. HOWEVER, the installed files where located in... "C:\Program Files"!!! The installer had somehow managed to install these files OUTSIDE the sandbox!!!
My virtual environment then ofcourse "ignored" these files as it had it's own copies of the Firefox files (v7) inside the virtual environment...
And I'm like... wha...?! I have no clue what's going on, and to make matters worse this problem was reproducable and the exact same for Thunderbird... anyone stumbled upon anything similar?
At one point I thought maybe %drive_C% had snuck in there with MERGED isolation modes, but not so far as I could see...
Anyone with ideas? Because I'm almost all out...