This week's meeting contains an excellent example of getting things done. In addition to routine issue triage and Rob's fortnightly WiX v4 status report, Jacob got an issue near and dear to his heart on the agenda, we talked it over, and signed Jacob up to do all the work. We should do this more often!
WiX v4.0 status
Today's WiX v4.0 status report was brief: Rob's continuing to make progress on the new pipeline model and compiler object model but the surgery is still ongoing and the patient is still on the table. And, not to overly mix metaphors, but as one of the midwives to this work, I can't wait to show it off.
Issue triage
Bundle is not registered in ARP during install if bundle is already cached, from @0ptikGhost, reports that a mildly hacked package cache results in a reinstalled bundle not getting registered in ARP. It isn't a problem that results from normal use but we agreed Burn should be more resilient to an unusual situation like this. Therefore, we took the issue into WiX v4.0 for investigation.
Overwrite="no" not working for certificate element, from @kennethgracilla, reports that the
Certificate
element'sOverwrite
attribute doesn't work. A bit of spelunking in the custom action code shows that the attribute is supported at build time but has no effect at runtime. We took the issue to show a warning in WiX v3.14, after which someone can investigate what's needed to make it functional in WiX v4.0.Unable to build when reference multi-targeting project, from @iskiselev, reports a problem using .wixproj projects referenced from other, "SDK-style" projects new in MSBuild v15. Rob volunteered to investigate.
Microsoft.Build.Utilities version 2.0.0.0 still referenced, from @ma-bjo, points out that the WiX MSBuild tasks still require .NET 3.5. That limitation has been removed already in WiX v4.0 but still needs to be addressed in WiX v3.14. Rob volunteered to make that happen.
Issue review
Jacob asked on the wix-devs mailing list about accessing variables from a bundle being upgraded. It sounded familiar, so I dug up the long-ago-opened issue: Allow access to persisted variables from related bundles. We discussed ways of accomplishing this, reached a general consensus on the best approach(es), and Jacob signed up to write a quick WIP.