Jorge wrote:
On Tue, Sep 02, 2008 at 10:30:40PM +0000, corvid wrote:
Currently, there's a MSG that tells you what encoding a form was submitted as -- once it's submitting it. Obviously that's of limited usefulness when you want to know beforehand what will happen to any Japanese that you might type in, but on other hand, so many pages have forms that MSGs ahead of time would be annoying, besides the fact that a page may have multiple forms with different accept-charset attrs...
Add to that the fact that one has to learn the hard way that you should only paste utf-8 into dillo2 and not, say, windows-1257. (and latin-1 works because fltk assumes it's latin-1 and does the conversion)
So my quick little experiment: text/password/textarea form controls display a tooltip that reads
enter: UTF-8
when UTF-8 will be submitted, and
enter: UTF-8 send as: [some other encoding]
otherwise.
That's a good start...
And it would clear things up for every person who tried to use google.com and wondered why non-latin-1 wasn't working when the oh-so-glorious dillo2 is supposed to be all UTF-8ed up. After all, it works in Firefox, right? (I've read that google pulls this idiocy with all of the little browsers. w3m, elinks, etc. Ignores the Accept-Charset header and does what it wants based on the User-Agent.)
It may be a but more complicated. Non-technical users may be left wondering what to do with that message or what UTF-8 is. For instance, I don't know how to input latin1 from an X environment that's set to utf8 (besides switching the whole thing to latin1).
AFAIU, which is not much in this regard, if there's a way to know what encoding the X server is using, dillo may iconv it into utf8 and then to the "send-charset" if necessary (in a similar way to what FLTK2 does with latin1).
That way it works for the user without further steps. Of course I may be wrong in this appreciation!
Would you mind explaining it further?
It deserves at least an entry in the FAQ (which I would read with interest, certainly ;).
I had been looking at this from the perspective of: "even though copy and paste of latin2 between two rxvt's works, I can't paste latin2 into dillo", but I suppose I'm not quite sure that X wouldn't have enough magic in it for latin2 to work for someone properly set up in a latin2 locale. Looking at the format of the Compose file, I _think_ it still wouldn't work for them (<Multi_key> <L> <minus> becomes "\243", not "Lstroke". Of course they don't need the Compose mapping, but the fact that it's going toward the octal encoding rather than a character seems significant). Putting aside what character encoding to enter, not knowing what encoding a form will be submitted as bugs me. Somewhat related to all of this: local files that are text/plain.