Ed Hurst <[EMAIL PROTECTED]> writes:
> I was wondering if there has been any progress toward a problem about
> which I asked a couple of years ago: ghost text. That is, on some
> webpages, parts of the rendered text will persist when scrolling up or
> down the page, repeating in rows.
Generally this happens when ELinks' idea of what is in the screen
gets out of sync with reality, most likely because the width of a
character doesn't match ELinks' internal unicode_to_cell().
(ELinks does not use wcwidth(), because that depends on LC_CTYPE
and POSIX does not specify the name of any locale where wcwidth()
would support arbitrary Unicode characters.) In ELinks 0.12.GIT,
I think there are three situations in which that can occur.
(1) Bug 824: If the document uses Unicode combining characters.
ELinks does not yet support those and they are not trivial to
implement.
(2) Bug 932: If the document uses C1 control characters (U+0080
to U+009F), UTF-8 I/O is disabled, and the terminal uses a
charset that includes those characters. Currently some of
ELinks' ISO-8859-* definitions include those but others
don't. In principle, all ISO-8859-* charsets should include
those characters, and ECMAScript code can test that. Instead
they should be filtered during rendering or terminal output
(isscreensafe). But e.g. CP437 has printable characters
in the 0x80...0x9F range so we should pass the charset info
to the filter.
(3) If the terminal assigns unexpected widths to line drawing
characters, like the Mac OS X Terminal does. There is not
much ELinks can do about this. In principle it could ask the
terminal to report the cursor position whenever it displays a
new character, but some terminals may ignore such requests,
and anyway it would be difficult to design dialog frames to
cope with random variation in character width.
If you find some other case, please report the document, the
terminal, and the terminal options you use in ELinks.
_______________________________________________
elinks-users mailing list
[email protected]
http://linuxfromscratch.org/mailman/listinfo/elinks-users
