Description of problem:
The anaconda text interface breaks if the U+201D RIGHT DOUBLE QUOTATION MARK characters - ” ” - are used in a translation. The problem is probably best demonstrated with a couple of screenshots:
Version-Release number of selected component (if applicable):
The anaconda version in Fedora 10 Beta
Always, though might be a bit difficult for non-Finnish speakers to reproduce
Steps to Reproduce:
1. Boot the Fedora 10 Beta DVD with lang=fi (as in Finnish)
2. In the first anaconda text interface screen, select OK
3. You should see something like http://vpv.fedorapeople.org/misc/anaconda-quotes-problem1.png
4. If you select "Tarkasta" in that screen, you should see something like http://vpv.fedorapeople.org/misc/anaconda-quotes-problem2.png
There's a box (that's probably <E2>) and <80><9D> where the character ” should be and the text "flows out" of the screen.
” (right double quotation mark) characters in the text interface should just work.
I've picked Severity: High, because this bug makes it quite difficult to use the text interface in Finnish.
I see two options for making the Finnish translations work in time for Fedora 10. Either the core cause of this bug (no Unicode support in the text interface?) gets fixed, or the Finnish translators have to convert all the right double quotation marks in the text interface into quotation marks (22).
I'd rather use the right double quotation marks in the Finnish translation, because that's what is generally recommended to be used when writing Finnish. In Anaconda's graphical interface they work fine.
This is caused by missing locale support, not newt's fault.
Slang doesn't print the ” character correctly because it's not in the UTF8 mode as nl_langinfo(CODESET) doesn't return "UTF-8". And the newt windows are broken because mbrtowc() and wcwidth() don't work.
Per Chris Lumens, this isn't a blocker.
*** Bug 469542 has been marked as a duplicate of this bug. ***
Our options for fixing this are:
(1) Bring wlite back into anaconda.
(2) Add all the glibc locale data into the second stage.
Both of these are horrible.
*** Bug 470927 has been marked as a duplicate of this bug. ***
This bug appears to have been reported against 'rawhide' during the Fedora 10 development cycle.
Changing version to '10'.
More information and reason for this action is here:
I'm still seeing this in Rawhide. Any chance of this being fixed before Fedora 11? Or should we start making changes to the po files to only use character which anaconda can deal with?
A lot of the text mode interface code is going to be going away before F11, so it's worth seeing if these strings will still even be present in text mode before making the drastic changes I outlined above. However if you do want to be absolutely sure, I'd start modifying the po files. It is very unlikely we will do either of my previous options.
For other translators following this bug report, here's a way of finding out how many bytes a character needs in UTF-8 (thanks to another Finnish translator for the hint): $ echo -n a | wc --bytes
It seems anaconda's text interface can deal with 1-byte and 2-byte characters, but not with 3-byte characters. With Finnish the situation is quite simple, since characters such as 'ä', 'ö' and 'å' are all 2 bytes, the character causing the problems is the 3-byte '”', which can be in this situation changed to '"' - even though '”' is generally recommended to be used when writing Finnish.
*** This bug has been marked as a duplicate of bug 492999 ***