`emptyUpdate` got new semantics with opaleye-0.4

[bergmark]

To fix this I think we should scrap runUpdateConst and replace emptyUpdate with another TH-generated function IdUpdate :: To Column R -> To Maybe (To Column R) which could be used like this:

runUpdate table (\r -> (f r) { name = Just $ constant x }) condition

This will make sure that all code that uses it has to get updated.

[hesselink]

What changed exactly about the semantics of emptyUpdate?

[bergmark]

tomjaguarpaw/haskell-opaleye#92

[hesselink]

What would runUpdate do? Would it set all columns explicitly to their old values? I think that might be tricky, since they might have been updated, like I mentioned in the opaleye issue.

[c-lewis]

Is this issue the reason why the Opaleye dependency has not been changed to allow 0.4?

[bergmark]

Yep, Everything else should work with 0.4. My suggestion for opaleye is here. I forgot to ping tom again, I'll do that now.