(Had to do it.)
It's been a week now since I started running Ganriki.org on Mercury. What's happened since then?
I've been reworking how the publishing queue activates when a page is published manually. The queue now runs in a separate tab and sends its results back to the page that invoked it, and will automatically close itself (and optionally pop up a desktop notification) once the queue run is finished.
I also fixed what appears to have been a small bug in the way the scheduled publishing script wasn't setting objects to "published" in the proper order.
To do: I still have to rewrite some of the way things are inserted into the queue to begin with. The way templates are manually republished, for instance, doesn't use a queued mechanism, and so there's potentially this long delay before the queue actually activates. So that'll be one of the things to fix next.
I've refined some of the code available to templates -- what are called the "tags" (not to be confused with tags assigned to individual pages in a blog), and which contain properties like the blog title, various helper objects, and so on. For instance, the
pages object now has properties like
pages.published. This makes it easier and more intuitive to get access to such things. I've also refined somewhat how code modules are made available in themes.
To do: I'm going to be adding a way to access a theme's code modules from within a template mapping, so that more complex mappings can be accomplished by way of simply using functions defined at the theme level.
This is a tricky one to get right, but I did implement some speedups by way of consolidating how pages are inserted into the queue to take better advantage of some existing caching.
To do: Nothing immediately but I want to keep my eye on how to make better use of caching for publishing runs.
All in all, not a bad week. Things will be slow for the next couple of weeks as I deal with real life, but running Ganriki.org on Mercury is giving me no end of opportunity to make both of them better.