Even if this is only a problem now in getting the right font, the box for the Cyrillic appears far too wide.

The user confirms that the lack of labels is fixed with the new release, provided he uses MS Arial Unicode to pick up the Japaneses glyphs. This still leaves the extra wide box.

[duerst] Could it be that somewhere bytes are counted instead of characters (which would just lead to a factor of 2)?

[ellson] I'm trying to work out the right way to deal with this.

Can you tell me what character encoding you are using to put Russian and Japanese text into the graph?

This information may come from some sort of localization data. Do you happen to know if there is an environment variable that dot should be looking at to find out what character encoding is being used?

On linux at least, you can tell what character encoding is being used with the command: locale charmap

Can you let me know what this returns on your system?

[duerst] The data is encoded as UTF-8. I know this because I produce it in a Java servlet, where I set the encoding explicitly, and I have also checked the actual encoding with od -hc (octal dump unix command), the Unicode book, and some of my personal utilities.

Please note that 'locale charmap' will only coincide with the actual encoding of the input file if the input file was created with the same setting, and by a program that uses the locale settings for deciding which encoding to use for the output. For data from the Web, for example, that doesn't apply.
Owner: ellson
Status: Fixed