That title is probably enough to put off most people from reading this post, but it might be enough to convince someone searching for a solution to their problem that this post can help them. If you are impatient, skip to the end where I give the one-liner change, but for the rest of you who want to understand the context, read on.
I have a [Debian] box which I had set up to use a United Kingdom, British English locale at installation. My KDE Control Centre [sic] said “Country or region: United Kingdom” and “Languages: British English” (after installing the requisite package). I had also used the locales package to create an en_GB.UTF-8 locale and set it as my default.
For quite some time, then, I had been happily using that system, assuming everything was working fine. Every now and then, though, I would check the character encoding of a document I had opened or created in Kate and seen it had defaulted to “Western European (ISO 8859-1)”. Similarly when saving or opening a file, it would say that the directory listing was in ISO 8859-1 encoding. For most files the difference was moot, and I was even prepared to believe this was some transient issue, but recently it started to annoy me.
That is, I admit, one of the hardest things about using Linux: not that things don’t always go how you want, but that when something doesn’t go how you want, the only thing stopping you fixing it is your own desire to learn. In a closed source environment, you can blame the company that made the software for putting artificial limits on the program (“We didn’t see a big enough market for feature X”, “The content industry won’t partner with us if we include feature Y”, “You have to upgrade if you want feature Z”) but with Linux you know there is a way, and other people out there have probably already done what you want, because they know how.
I’m not saying Linux is full of these experiences, in fact, if I didn’t have such a high expectation for what it could do, I probably wouldn’t face these challenges at all. After all, how do you make Microsoft Word default to UTF-8 encoding of documents, or a Windows desktop use UTF-8 as the character encoding for directory listings?
Reminding myself how much I liked a challenge, I hit the Google-machine and started doing queries similar to the title of this post. The result was a collection of comments about environment variables like LANG, LANGUAGE and LC_CTYPE, and files like /etc/environment, /etc/locale.alias and /etc/locale.gen. This left me with a finite number of possible combinations for settings, which I could have tried individually before getting the setting I wanted. That method suffered from the fact that to test my settings, I would have to log off and log on, but was made easier by the existence of the locale command which “writes information about the current locale environment, or all locales, to standard output”.
Using this command, I spotted what was potentially the problem — it gave output like this:
LANG=
LANGUAGE=en_GB:en_US:en_GB:en
LC_CTYPE="en_GB.UTF-8"
LC_NUMERIC="en_GB.UTF-8"
…
with the LANG variable not being set. The method I used to change this was adding a second line to /etc/environment so that it said:
LANGUAGE="en_GB:en_US:en_GB:en"
LANG="en_GB.UTF-8"
Sure enough, the LANG variable was then set, and KDE behaved properly on the next login. A surprisingly simple change to make, but difficult to find.
Is there a method for connecting all pages that deal with a certain problem together, and then linking all pages which answer them together? Blogs with TrackBacks are fairly good for issues which are localised in one region of time, but what about perennial problems?
