Patching question

[blockda]

Hello, I'm sort of new to trying to get patching working on my application, here is a bit of background:

So my application is about 10000 files, in about 1300 components. It takes a while to install, so for the nightly development build, I wanted to make a small patch that wouldn't take long to install.

Now I have my application built, I backed up my built installer, upped the revision number from 1.0.0 to 1.0.1, added the backed up installer to the patch references. Now I add a file or two to my source files, and rebuild the package and patch. The patch is about 1.5MB, which is appropriate with included runtime files and patch data.

Now for the problematic part that I'm trying to get more information on. When I run the patch, it takes forever to apply because it goes through every file in the original distribution and evaluating weather or not it should patch the file. This makes my 1.5MB patch installer take more than 15-ish minutes to install.

Is there any way to instruct the setup process that it only needs to patch a few files, so it doesn't spend a whole lot of time checking things that don't need to be updated?

[blockda]

I think I found some more information about this topic.

I was browsing around the msdn docs, and found that what I think I am searching for is something called "Patch Optimization"

http://msdn2.microsoft.com/en-us/library/aa370595.aspx

Now, there is a bit in that document that says in windows installer land after version 3.1, there must be the "OptimizeInstallMode" property set to 1 in the PatchMetaData table.

I fished out the .msp file from my patch by letting the installer run and opening up the %temp% area where the files were extracted to. I opened the .msp file in orca and found that there is no PatchMetaData table in this patch.

Which leaves me with 2 questions:
1) Is patch optimization (e.g. small patch and fast installation time) possible in InstallAware?

2) Is it possible to generate a patch that is uninstallable with InstallAware? (this is because apparantly the ARP data is pulled from the PatchMetaData table)