Gedit Review (Softpedia.com) [LWN.net]

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 22:55 UTC (Mon) by khim (subscriber, #9252)
In reply to: Gedit Review (Softpedia.com) by sfeam
Parent article: Gedit Review (Softpedia.com)

Last time I've checked gEdit was the only editor with correct handling of UTF-8. Emacs ? Kate ? Don't make me laugh.

Do a quick check:
$ printf '\x61\xCC\x8A\xCC\xA7\xCC\x81\xCC\x81\n' > test.txt
$ gedit test.txt &
$ emacs test.txt &
$ kate test.txt &
gEdit shows text correctly (single character "a with a ring above, cedila below and acute accent over all this") while kate and emacs show... something. Not even remotely close to what I want to see...

Yes, gEdit is memory hog and it's stability is horrible, but at least it can read UTF-8 text - that's more then I can say about VIM, Emacs or Kate (but on the other hand may be I just don't know how to tune them correctly? feel free to explain what I did wrong - I've used versions from last version of Kubuntu).

to post comments

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 23:11 UTC (Mon) by nix (subscriber, #2304) [Link] (11 responses)

Emacs is *definitely* capable of UTF-8 text display (and text in virtually any other encoding one could possibly imagine).

There must be a configuration problem somewhere (probably in the X11 fonts, that's where problems tend to lie).

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 23:18 UTC (Mon) by khim (subscriber, #9252) [Link] (9 responses)

Emacs is *definitely* capable of UTF-8 text display (and text in virtually any other encoding one could possibly imagine).

Are you sure ? Can you see "'a' with a ring above, cedila below and acute accent over all this" character in emacs?

There must be a configuration problem somewhere (probably in the X11 fonts, that's where problems tend to lie).

May be - but then why gEdit can show text correctly ? As I've said I would not be surprised is emacs has some mode where it can show combining characters right (emacs is wast and there are a lot of knobs) but I was unable to find such mode :-(

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 3:08 UTC (Tue) by aseigo (guest, #18394) [Link] (7 responses)

it's very simple. if your env is not set to read/save "utf8" then katepart
doesn't assume that's the encoding for the file. in which case go to the
open file dialog and ensure that utf8 is the selected encoding in the
upper right corner.

but katepart does indeed handle utf8, and dozens or other encodings,
absolutely perfectly.

Encoding in Kate

Posted Jun 13, 2006 6:29 UTC (Tue) by man_ls (guest, #15091) [Link] (6 responses)

I have often wondered why doesn't kwrite auto-detect text encoding? It should be fairly easy to do, but every time it pretends that my ISO 8859-15 encoded text is UTF-8. Even when it's an HTML file and the webpage says otherwise. It is a bit annoying.

Encoding in Kate

Posted Jun 13, 2006 9:12 UTC (Tue) by halla (subscriber, #14185) [Link] (5 responses)

Autodetecting a text encoding is definitely not simple at all -- in fact,
most people who've tried it agree that it's impossible do do correctly
and reliably.

Encoding in Kate

Posted Jun 13, 2006 9:16 UTC (Tue) by piman (guest, #8957) [Link] (2 responses)

Autodetecting UTF-8 is pretty easy, since very few valid sequences are valid UTF-8. So UTF-8, fallback to locale encoding, is very effective. If a UTF-8 or UTF-16 BOM is present, it's also almost always right (though ugly).

When downloading web pages, detecting the encoding is very easy, the server gives it (if the server is wrong, it's not your editor's fault, and it will be right more often than not).

It's only when forced to choose between old legacy encodings (Latin 1 versus Latin 2, for example) that it becomes difficult-to-impossible to guess the encoding.

The final good solution is to assume everything is UTF-8 unless otherwise specified and force the last legacy users to migrate to something sane. This is often best on a multiuser system where LC_CTYPE won't apply outside the current user's /home resources.

Encoding in Kate

Posted Jun 13, 2006 14:29 UTC (Tue) by tjc (guest, #137) [Link]

The final good solution is to assume everything is UTF-8 unless otherwise specified and force the last legacy users to migrate to something sane.

Sounds a bit Orwellian to me. Even if you mostly succeed, a few people will still meet in abandoned building at the dead of night and read ASCII text...

Encoding in Kate

Posted Jun 13, 2006 14:44 UTC (Tue) by man_ls (guest, #15091) [Link]

People will use whatever works best for them. For me, it's latin-1 or ISO 8859-15. It is not a legacy encoding; it is completely sound if all you want is to write European characters. I'm fairly sure I will never mix any Chinese text in my web pages, and with ISO 8859-15 they seem to work best across editors and browsers.

This web page is written in ISO 8859-1; or so it says in the header:

CONTENT="text/html; charset=iso-8859-1"

so the character might not work. In fact it does work in Firefox, maybe because UTF-8 is really the charset used?

Encoding in Kate

Posted Jun 13, 2006 9:25 UTC (Tue) by man_ls (guest, #15091) [Link] (1 responses)

Note that I'm not suggesting to autodetect the encoding blindly, but based on the XML header:

<?xml version="1.0" encoding="ISO-8859-15"?>

that appears as the first line of an XHTML file. Maybe the choice of word "autodetect" was unfortunate. The encoding in an XML document is declared and should be respected, right?

Encoding in Kate

Posted Jun 14, 2006 11:28 UTC (Wed) by niner (guest, #26151) [Link]

But kate part is no XML editor. It may have some plugins making work with XML files easier, but it's still a normal text editor that knows nothing about XML.

Gedit Review (Softpedia.com)

Posted Jun 16, 2006 20:53 UTC (Fri) by vonbrand (guest, #4458) [Link]

Nope, here I see: 'a', 'o', and two '. With gedit, less, etc. xemacs and emacs show 'a???' and 'a<3 boxes>', respectively.

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 19:58 UTC (Tue) by nim-nim (subscriber, #34454) [Link]

Gee, right the problem is probably in the X11 core fonts backend emacs insists in using years after everyone switched to fontconfig.

This *still* makes emacs guilty in my book. If an app purposefully chose not to switch from a broken backen, it's as guilty as if the brokeness where in its own code.

If there were no alternative to core fonts, now you could blame the backend.

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 23:15 UTC (Mon) by busterb (subscriber, #560) [Link] (1 responses)

Yeah, I think that the UTF-8 support is largely dependent on the distro
and libraries being configured correctly more than the editor these days.
On a relatively-untweaked Kubuntu install:

Gedit displayed 5 characters, the a, the ring, the cedila and then two
grave accents.

When I copied from gedit and pasted into this edit window, I got the
correct single character.

Kate had no probems displaying the single character.

Vim displayed a single character with a ring, but no cedila.

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 23:26 UTC (Mon) by khim (subscriber, #9252) [Link]

1. Select "Arial" font in gEdit.

2. Strange - I was unable to make Kate show single character :-( May be I should select different font ?

And yes - VIM's problem are mostly terminal-related...

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 23:18 UTC (Mon) by jwb (guest, #15467) [Link]

vim actually displays this correctly. Perhaps your terminal emulator isn't quite up to the task.

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 23:21 UTC (Mon) by arafel (subscriber, #18557) [Link] (5 responses)

Whereas on my machine (Kubuntu Breezy), GVim is the only one which shows that
correctly, with Kate a second, and GEdit more or less failing. (It shows the characters
individually, rather than composed.)

I think this is one of those cases where "your distro may vary".

Gedit Review (Softpedia.com)

Posted Jun 12, 2006 23:29 UTC (Mon) by khim (subscriber, #9252) [Link] (4 responses)

gEdit needs variable-length font to work correctly - that's not the case for all distributions. Just select Arial and you should be all set.

P.S. The strange thing is that "fix" was suggested by my friend who's using Microsoft Word and knows nothing about gEdit, VIM or Kate :-)

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 0:01 UTC (Tue) by arafel (subscriber, #18557) [Link] (1 responses)

It's better, but still not right - I get a-acute now. :)

This is one thing I still dislike about desktops under Linux - if you get the right fonts installed (both in terms of charset and viewability) it looks really nice, but sometimes you have to go hunting.

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 18:52 UTC (Tue) by oak (guest, #2786) [Link]

This might be useful: http://www.unifont.org/

Gedit Review (Softpedia.com)

Posted Jun 14, 2006 9:01 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

http://dejavu.sf.net/

Gedit Review (Softpedia.com)

Posted Jun 14, 2006 9:57 UTC (Wed) by micampe (guest, #4384) [Link]

gEdit needs variable-length font to work correctly - that's not the case for all distributions. Just select Arial and you should be all set.

Not really, I use Terminus and I see your character just fine.

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 2:41 UTC (Tue) by Per_Bothner (subscriber, #7375) [Link]

I tried JEmacs running on Sun's JDK 1.4.x running on Fedora Core 5 and it actually worked - it displayed a single composed character. I take no credit - the Swing code just seems to do the right thing. (And I certainly don't claim that JEmacs is ready for serious work - it crashes very easily. It's mainly a proof-of-concept at this point - and likely to remain so unless someone gives it some TLC.)

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 6:12 UTC (Tue) by halla (subscriber, #14185) [Link] (1 responses)

If Gedit assumes that a file is utf8 when LC_CTYPE is not set to something.UTF-8 (like
LC_CTYPE=en_US.UTF-8) looks like seriously broken behaviour to me. And if LC_CTYPE is set
correctly (as it should have been set by your distribution), then Kate will certainly show your text
correctly, konsole will show your text correctly if you cat it, vim will show your text correctly --
in fact, only Nedit might have trouble. Unless Nedit improved a lot in the last year.

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 14:45 UTC (Tue) by jpetso (subscriber, #36230) [Link]

> If Gedit assumes that a file is utf8 when LC_CTYPE is not set to
> something.UTF-8 (like LC_CTYPE=en_US.UTF-8) looks like seriously
> broken behaviour to me.

Well, but it's a fact that not every file corresponds to your local
LC_CTYPE settings. For example, my newer text files (since I switched to
Kubuntu) are encoded in UTF-8, and the ones from my Gentoo age and before
still are ISO-8895-1. Also, everything that involves other people can't
be fixed to the one correct encoding. KDE svn might use UTF-8 files while
group projects at university courses often involve ISO-8895-1 ones.
Also, web pages are quite differently encoded, even if XHTML pages
specify their own encoding.

Opening text files based on an environment variable is broken by design.
Only new files should be created with regard to LC_CTYPE, whereas
existing ones should be checked on their encoding.

Gedit Review (Softpedia.com)

Posted Jun 13, 2006 11:49 UTC (Tue) by debacle (subscriber, #7114) [Link]

I'm using both GNU Emacs and (sometimes) vim on Debian etch and both work absolutely fine with UTF-8. I never tried kate.

Gedit Review (Softpedia.com)

Posted Jun 17, 2006 15:48 UTC (Sat) by kreutzm (guest, #4700) [Link]

Well, my gvim is showing this letter fine once I enable UTF8 (e.g. in the config file or on the :prompt): set enc=UTF8

Emacs I don't know.