if you see me coming, run.
About all the demographics templates soap operaModificar
First of all, let me congratulate Occitan Wikipedia community on great work and also on all the great soap and shampoo. The reason of me being here is not pleasant neither for me, nor for anyone else. Occitan Wikipedia 'Demografia' and few other templates relied on the idea, that templates can be used as databases, which was already exposed as wrong one back in the day, when someone decided to build Australian Rail Station Database using this method. Apart from being not designed for this task entirely, templates storing data remove that data from the article context, kind of destroying various reusability cases for wikis.
In this case, there were twenty or more megatemplates (each having tens of thousands of #switch statements) included to show some small snippet in the article. Rendering such articles was impossible (I wonder, did anyone realize, that any pages using those templates never rendered?) - yet they were consuming plenty of resources on their way to failure. Generally, an attempt to render a page that had one of those templates, used nearly a gigabyte of RAM, until it failed, but in some cases that memory usage had detrimental effects on whole website, including e.g. Commons and English Wikipedia, and everyone else. Even worse, sometimes that would go to all job queue runners, and blow up in memory on all at once.
In my first deletion I wrote, "this breaks the site". It does. It doesn't work for you, it doesn't work for anyone, and it has very bad effects. Pretty much every fix we will be coming up with will break these templates.
- Large #switch statements will be constrained to #100 or so, and we will probably have to work more on fixing recursion in such cases. This will break these templates.
- Memory usage will be properly constrained, once we move XML-DOM module memory allocations under the PHP's memory management. This will break these templates.
- Pretty much anything we will do will break these templates. There are some discussions about how to embed data from other sources into wiki pages, and someone is actually working on it, but I don't trust things like that too much :-) A better wiki-side scripting language is needed for that, so the technology shift is much larger than anyone really thinks of.
Therefore, I'd like to point out, that we will have to delete those templates one way or another, and there's not much of compromise that can be reached. Meanwhile, there's quite a bit of hilarity in all this process. The idea of 'global block' is hilarious by itself, Ting Chen not recognizing me is hilarious too, he probably missed his morning coffee. For now I will go have breakfast, some more morning coffee and work on my daily job stuff, will see what happens afterwards.
English Wikipedia community has learned certain things by now: http://en.wikipedia.org/wiki/Wikipedia:PERF#If_the_sysadmins_identify_a_performance_problem.2C_they_will_fix_it
P.S. I know that I could've identified myself properly and written more explanation, but there was already lots of explanation in my first deletion message.
This community is apparently has some sick people: https://en.wikipedia.org/w/index.php?title=Wikipedia:Administrators%27_noticeboard&oldid=716663750#Admin.28.3F.29_Midom_causing_disruption_on_Occitan_Wikipedia