On Saturday, October 2, 2010, Loui Chang <louipc....@gmail.com> wrote: > On Fri 01 Oct 2010 21:41 -0500, Dan McGee wrote: >> On Fri, Oct 1, 2010 at 9:39 PM, Loui Chang <louipc....@gmail.com> wrote: >> > On Thu 30 Sep 2010 18:19 +0200, Lukas Fleischer wrote: >> > I was thinking - could we just change things so that OutOfDateTS is the >> > sole indicator of whether a package has been flagged or not? >> > >> > If it's set, then it's out of date. >> > If it's zero or null, then it's up to date. >> > >> > This might also apply to deletion, or orphan requests. >> > Really all these types of flagging could share the same structure and >> > code as far as I can envision it. >> >> I made this exact change to the archweb code a while back; this is a >> great usage of a nullable column. >> http://projects.archlinux.org/archweb.git/diff/?id=edfdc2&id2=845838 > > Yep. It makes sense to me. > >> Skip over the new migrations code (main/migrations/*.py); and look at >> the other changes if you are curious. > > Wow that migrations code is crazy. > I just added a text file that people can read and manually update their > database.
Crazy if you're viewing it the way you are as something to try and understand. 100% awesome if you realize the generation and application of them is completely automated. Trust me, it is a huge step up and the additional complexity makes it a lot more flexible and robust. -Dan