(I subscribed again, and this mail originally sent by Mike to me I now send to the list. My old e-mail address will no longer be in use.) Van: Mike FABIAN [mailto:mfabian@suse.de] Verzonden: wo 5-10-2005 13:56 Aan: hermanmeester CC: Onderwerp: Re: culmus font package available for SuSE Linux 10.0
Answering yor question: I'm missing, or haven't found yet, a way to type Greek accents (most important, the aspiration sign) and the "iota-subscriptum" beneath the Greek omega and �ta. I think Greek might need a feature similar to M17N-t-latn-pre, for example. I guess Linux users from Greece may already use this. (In fact, I'm only processing some crazy apocalyptic Greek writings from 200 BCE, when they didn't even invent those accents yet. But some academics are so perfectionistic, they criticised my unaccented typing.)
Please have a look at the file /usr/share/m17n/el-kbd.mim I guess it is very easy to add the things you want to that file. I this file you find a table like this (shortened): (map (map ("q" ?\;) ("w" ??) [...] ("N" ??) ("M" ??) (";a" ??) (";e" ??) [...] (":;y" ??))) This means if you type a 'q' you will get a ';' inserted (the '\' is needed as a quoting character because ';' would otherwise start a comment in lisp). If you type 'w' a '?' will be inserted. Typing sequences of more than one character are also possible, i.e. if you type ';a' a '?' will be inserted, if you type ':;y' a '?' will be inserted. This table can be easily extended to support additional key-sequences. If you think you have suggestions how to extend that table which are of general interest, please edit this file an send a diff to the developer of m17n-lib (this is Kenichi Handa <handa@m17n.org>, there is also a mailing list, see http://www.m17n.org/m17n-lib/). If you want some input key-sequences which are very personal and maybe only interesting for you, you can easily create your own input method. Create a directory ~/.m17n copy a m17n input method file which is already a bit similar to what you want into that directory: cp /usr/share/m17n/el-kbd.mim ~/.m17n/herman.mim Edit this file, change title (title "?") -> (title "herman") edit the table as you like. Create a file called "mdb.dir" in the ~/.m17n directory. Put the following line into that file: (input-method t herman "herman.mim") This file is an index of input methods in the directory ~/.m17n. For each input method you add you need such a line. Now add export M17NDIR=$HOME/.m17n to your ~/.profile or ~/.bashrc. This variable tells the m17n library to look for additional input methods in ~/.m17n. Now restart your X session. Activate SCIM and your new input method called "herman" should show up under "other" in the SCIM input method menu.
As for Hebrew, I would love SCIM to enable vocals typing below, inside and above the Hebrew characters, but this may well be something I have to produce myself, because even the Israeli's I asked didn't know this for OpenSource yet. I've been taking a look at this with a university teacher in Leiden, who designed his own fonts occasionally. I'll first need some time to convert him to Linux, however ;)
With the above explanation, maybe you can add this feature to the m17n-lib yourself and send the result to the Mr. Kenichi Handa or the m17n-lib mailing list.
By the way Mike, I met a student of Japanese in Leiden, he was interested in Linux, but he said he could type anything he needed in Japanese using MSWindows, and didn't want to "change a winning team". Is there anything about SCIM or UIM or other methods in the field of Japanese typing that beats the Windows input method? (to "draw him over the line"?) I guess there must be, but I don't know any Japanese, nor did I use Windows for a few years.
The Japanese input method in windows is quite intelligent. The Japanese input methods available as free software on Linux are currently less intelligent, i.e. the user has to think more while typing and do more manual corrections. Japanese input is usually done by typing the phonetics in Latin alphabet. The input method converts this to Japanese characters. The main problem is, the this conversion is *not* unique. This is very different from something like M17N-t-latn-pre where '"u' is always converted to '�'. In Japanese, there are almost always many possibilities how to convert. Intelligent input methods manage to choose the right conversion from the context most of the time. Less intelligent input methods often get it wrong when converting and the user has to correct manually. For me this is not a big problem, the free input method I am currently using for Japanese (scim-anthy) gets it right maybe 95% of the time if I hit the conversion key at the right moments. One has to get a feeling where in a sentence it is good to hit the conversion key. If one has developed this feeling for an input method like scim-anthy, it also works quite well. But windows users are not used to that. They often type very long sentences without pausing and then hit the conversion key and expect that everything is converted correctly. Even with the best intelligent input methods this doesn't always work. But the commercial input methods can handle this typing style much better than the free input methods. Commercial input methods are available for Linux as well but have to be bought separately. And they are quite expensive. With the Japanese version of SuSE Linux 9.1, the commercial input method "ATOK X" was bundled. With the Japanese version of SuSE Linux 9.2 and 9.3, the commercial input method "Wnn 7" was bundled. For SuSE Linux 10.0, Novell Japan did choose not to bundle any commercial input method any more to cut costs. I.e. if a user wants to use "ATOK X" or "Wnn" on SuSE Linux 10.0, he has to by it directly from the vendors. I think that scim-anthy is really usable, it works reasonably well if the user knows the best places where to hit the conversion key. But as this is different from Windows, the user has to get used to that.
I do know that the SCIM menu, for KDE, looks a lot better than the MSWind. input method, which looks as if it was created in the '80s.
Mike, thanks a lot, that is very helpful, I'll get to work immediately! Regards, Herman -- Mike FABIAN <mfabian@suse.de> http://www.suse.de/~mfabian