UESPWiki:Administrator Noticeboard/Archives/Special Characters

This is an archive of past UESPWiki:Administrator Noticeboard discussions. Do not edit the contents of this page, except for maintenance such as updating links.

Special Characters

I notice that the edit window has these cute little JavaScript links to insert special characters into the text box. Very handy, of course, but may I put forth that this is ultimately a bad idea? I've never been an admin of a MediaWiki installation, but I've been a major contributor to two other MediaWikis (the Homestar Runner Wiki, and the Kingdom of Loathing Wiki), and both of them have seen firsthand the havoc that can be caused by putting non-ASCII characters directly into WikiCode.

The problem is subtle. This technique appears at first glance to be no problem, as long as the database, MW installation, and web server are configured appropriately. But I've seen the troll fat hit the fan when a Wiki is migrated to a different platform. Migrating from Windows to Linux, I believe, is a particular disaster. I couldn't tell you exactly what causes it, but what can happen is that all non-ASCII characters in page texts will be silently replaced with garbage or question marks. The cleanup of a spill like this, as you can imagine, is a nightmare, especially if your old installation is no longer accessible by the time you notice it.

On the other hand, non-ASCII characters that are entered as character references (e.g., é = é or é = é) never require the database, MW software, web server, migration script, or browser to understand any encodings but ASCII. Much more bulletproof. And they will still display just as nicely in the browser; possibly better, if the browser doesn't like the encodings that the server knows.

I recommend that a brief explanation and directive to use character references be added to the Format section of the UESPWiki:Style Guide. See the Standards page of the HRWiki for an example. And if possible, I strongly recommend that those quick-insert links be modified to insert the appropriate character references instead of real characters. That would make compliance with such guidelines much easier than it has been on other wikis. Thoughts? Does anyone know an argument against this, or reason to be positive it will never be an issue for this wiki? --TheNicestGuy 14:36, 10 January 2008 (EST)

Well, for one thing I'm not sure it's POSSIBLE to have those little links display something different from what actually appears in the code. It's not really Javascript so much as an obscure, modified form of wiki markup. (Though it's not actual wiki markup, so the usual techniques won't work there. You can look at the code here: MediaWiki:Edittools (though only Admins can edit it.) You'll notice that the biggest grand-daddy wiki of them all, Wikipedia, has a similar feature on their edit page, see the code here. They include far more in the way of accented characters and such to account for the various subjects likely to be discussed there. But you'll notice their code for creating these characters is pretty much identical to ours. If it's not a problem for them, I don't see why it should be for us. As for possibly migrating to a Linux server - I think we already are on one, but you'd have to ask the experts. At any rate, I don't anticipate that we're going to move to another OS any time soon. A faster server, maybe, but I imagine we'll continue to run the same software for quite some time. At any rate, changing all this would be such a major task that it's almost not worth the effort. --TheRealLurlock ^Talk 15:00, 10 January 2008 (EST)

I just tried to adjust the file to use proper HTML codes (eg, × for ×) and couldn't get it to work either. If the alternative is switching the Insert bar off, I'd say that it's better to live with what we've got even if it causes a problem. –Rpeh^{•T•C•E•} 11:31, 11 January 2008 (EST)