From: François P. <fra...@mi...> - 2014-01-12 23:05:09
Attachments:
signature.asc
|
Bonsoir, I have just compiled mlterm and I don't know how to use it with indic scripts. I run configure script like this: ./configure --enable-m17nlib --enable-ibus --with-type-engines=xft --enable-ssh2 --enable-ind And install mlterm without any problems. Now I am wondering how to run mlterm to display indic scripts: I can use iBus, but if I use harvard-tokyo input method through iBus, mlterm cannot display the devanagari script. How can I configure mlterm to use otf fonts like FreeSerif which is able to display both roman and devanagari? And is it possible to have many indian scripts at the same time: devanagari, tamil, etc. Thanks for any help. -- François Patte UFR de mathématiques et informatique Laboratoire CNRS MAP5, UMR 8145 Université Paris Descartes 45, rue des Saints Pères F-75270 Paris Cedex 06 Tél. +33 (0)1 8394 5849 http://www.math-info.univ-paris5.fr/~patte |
From: Mayuresh <may...@ac...> - 2014-01-13 03:26:01
|
On Sun, Jan 12, 2014 at 11:51:06PM +0100, Fran?ois Patte wrote: > Now I am wondering how to run mlterm to display indic scripts: I can use > iBus, but if I use harvard-tokyo input method through iBus, mlterm > cannot display the devanagari script. > > How can I configure mlterm to use otf fonts like FreeSerif which is able > to display both roman and devanagari? I have some observations about devanagari on mlterm 1. The command that worked for me is: mlterm --bi=false --ind -E utf8 -V --type=cairo -w 24 with ~/.mlterm/aafont containing ISCII_HINDI= DV\-TTYogesh 2. -E isciihindi is a valid argument, though it does not work. --type=cairo or xft, both work with some minor change in appearance. It is important to watch .mlterm/msg.log to see whether there were any rendering errors. 3. There are more popular fonts such as Lohit Marathi, which strangely do not work. I wonder whether the space in their name is a problem. 4. README.indic in the package is handy. 5. I found this working only on Linux. I have been trying to get mlterm work on NetBSD with no luck. It shows garbled fonts for exactly same build and CLI arguments. 6. Rendering is also a function of the application. E.g. if I cat a file on the terminal it renders better than it does the same file in vi. In a browser like elinks, some devanagari text was shown fine while some was not. If I cat a file through paging commands like more/less, it doesn't render it properly. 7. Lastly, the rendering is very approximate and has several limitations. It is better than some of the other terminals but inferior to what a web browser, office or even what a tool like pango-view shows. Mayuresh |
From: Mayuresh <may...@ac...> - 2014-01-14 03:43:25
|
On Tue, Jan 14, 2014 at 07:49:51AM +0900, Araki Ken wrote: > > 7. Lastly, the rendering is very approximate and has several limitations. > > It is better than some of the other terminals but inferior to what a web > > browser, office or even what a tool like pango-view shows. > > The support of indic scripts depends on libind library which was developed > 15 years ago. > > libind converts ISCII text to the glyph indexes of such fonts as DVYG0ntt.ttf > and DVSR0ntt.ttf which contain glyphs in a different order from modern unicode > fonts. > * The glyphs in DVG0ntt.ttf is http://mlterm.sf.net/ttyogesh-glyphs.png > * The conversion table from ISCII hindi text to the glyph indexes is > https://bitbucket.org/arakiken/mlterm/src/tip/libind/table/hindi.table > > I'm not familiar with indic scripting, but I think the current processing > depending on libind is obsolete. > I would like to change it to the modern and light-weight alternative > if available. > > I'll be happy if you report strange behaviors or bugs of libind conversion > tables etc in detail or suggest alternative libraries. I am really enthused by your mail. I have been desperately looking for a devanagari terminal for ages now and asked nearly every possible mailing list and contact. If there is anything I can do to improve mlterm's support for indic, I'll be glad. I'll take a look at above links, though have a basic question. How does a tool like pango-view or firefox render devanagari correctly? Is it possible to adopt what they do? Would also like to understand the application's role in rendering. E.g. if I cat a file on the terminal it renders alright, but same in vi/more/less does not look the same, not in elinks either. mutt, elinks and vim are the 3 applications that matter to my day to day usage a lot. Mayuresh. |
From: Mayuresh <may...@ac...> - 2014-01-17 17:53:44
|
On Sat, Jan 18, 2014 at 02:14:56AM +0900, Araki Ken wrote: > If you know some examples of console applications rendering indic scripts > correctly on a terminal, please inform me of them. I can give an example of elinks. As of mlterm I am able to run, it shows only some of the letters on this page in devnagari and rest all as asterisks. Let me know whether you see full page in devnagari. http://maharashtratimes.indiatimes.com/newsletter.cms I have another question with X11 itself. If I select devnagari text from firefox and paste it in mlterm, it just shows ????. What could be the reason? > BTW, I found some mlterm bugs of rendering indic scripts. > Mainly, mlterm converts unicode indic scripts to ISCII ones internally, but > U+950 - U+95e weren't correctly converted to ISCII because each of these characters > is equivalent to *two* ISCII characters. (Some unicode characters (e.g. U+960 - > U+963 etc) which I couldn't find equivalent ISCII characters are still untouched.) Can you elaborate this a little bit. If it is a script related question, if you can give me some steps, I can definitely help. > I fixed these bugs, so try hg head. > => https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz Will try. > The difference between 3.3.2 and hg head is http://mlterm.sf.net/mlterm-fixindic.png I saw the screenshot. That is a remarkable improvement. Hope we get it in the next release soon. (I'll try head for now.) > PS: If you find text files which pango-view and mlterm show differently, please > send it to me. I can say pango-view and firefox render devnagari nearly perfectly. So if you are not familiar with the script and see a garbled character I am nearly sure you can trust the rendering of these two. I could cite a lot of text that was correct with pango-view and incorrect with mlterm. But after seeing your screenshot, I feel a lot of it might be taken care of. I'll try with head and let you know if I find discrepancies. Mayuresh |
From: Araki K. <ara...@us...> - 2014-01-24 14:50:55
|
Hi, From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Fri, 17 Jan 2014 23:23:37 +0530 Message-ID: <201...@SD...> > I have another question with X11 itself. If I select devnagari text from > firefox and paste it in mlterm, it just shows ????. What could be the > reason? I can copy&paste devanagri text from firefox (bundled in ubuntu 13.10) to mlterm. Can you copy&paste it from firefox to gedit? And, what happens if you start mlterm with --viaucs option ? --- Araki Ken ara...@us... |
From: Mayuresh <may...@ac...> - 2014-01-18 08:23:25
|
On Sat, Jan 18, 2014 at 02:14:56AM +0900, Araki Ken wrote: > I fixed these bugs, so try hg head. > => https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz I get the following error when trying to compile the above on tiny core linux. Any leads? libtool: link: `x_font_xft.lo' is not a valid libtool object make[1]: *** [libtype_xft.la] Error 1 make: *** [all] Error 2 Mayuresh. |
From: Mayuresh <may...@ac...> - 2014-01-15 15:32:33
|
On Tue, Jan 14, 2014 at 07:49:51AM +0900, Araki Ken wrote: > libind converts ISCII text to the glyph indexes of such fonts as DVYG0ntt.ttf > and DVSR0ntt.ttf which contain glyphs in a different order from modern unicode > fonts. > * The glyphs in DVG0ntt.ttf is http://mlterm.sf.net/ttyogesh-glyphs.png > * The conversion table from ISCII hindi text to the glyph indexes is > https://bitbucket.org/arakiken/mlterm/src/tip/libind/table/hindi.table > > I'm not familiar with indic scripting, but I think the current processing > depending on libind is obsolete. > I would like to change it to the modern and light-weight alternative > if available. > > I'll be happy if you report strange behaviors or bugs of libind conversion > tables etc in detail or suggest alternative libraries. I am not sure how to check the tables. What are the rectangular boxes in the glyphs? I see those often in mlterm. Many of them could be incorrect. http://mlterm.sf.net/ttyogesh-glyphs.png There are some other glyph based fonts, which look cleaner. I'll try these and provide feedback whether they do better than ttyogesh in mlterm. https://www.typotheque.com/fonts/fedra_hindi/book/glyphs BTW Would looking at this code used by lexilogos help you in any way? http://www.lexilogos.com/clavier/cardeva.js Mayuresh |
From: Araki K. <ara...@us...> - 2014-01-24 14:50:42
|
Hi, From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Wed, 15 Jan 2014 21:02:25 +0530 Message-ID: <201...@SD...> > I am not sure how to check the tables. What are the rectangular boxes in > the glyphs? I see those often in mlterm. Many of them could be incorrect. > http://mlterm.sf.net/ttyogesh-glyphs.png Mlterm converts unicode characters to iscii ones by mkf/libtbl/mkf_ucs4_iscii.c, then converts them to glyph indexes of ttyogesh font by libind/table/hindi.table. I think mlterm sometimes shows such rectangular boxes because mlterm parses indic scripts wrongly. I partially fixed this problem of devanagari, but haven't solved yet completely. Moreover, it is still very buggy to show indic scripts except devanagari. If you notice a word containing rectangular boxes or something strange, please copy and send it to me. Regards, --- Araki Ken ara...@us... |
From: Mayuresh <may...@ac...> - 2014-01-15 15:47:00
|
On Tue, Jan 14, 2014 at 07:49:51AM +0900, Araki Ken wrote: > I'll be happy if you report strange behaviors or bugs of libind conversion > tables etc in detail or suggest alternative libraries. One of the biggest problems is - vim not being able to render the characters the way cat renders on mlterm. What could be the reason? Mayuresh. |
From: Araki K. <ara...@us...> - 2014-01-24 14:50:47
|
Hi, From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Wed, 15 Jan 2014 21:16:53 +0530 Message-ID: <201...@SD...> > One of the biggest problems is - vim not being able to render the > characters the way cat renders on mlterm. What could be the reason? Now vim on my Ubuntu 13.10 shows indic characters as follows. http://mlterm.sf.net/mlterm-indic-vim.png The version of vim is: VIM - Vi IMproved 7.4 (2013 Aug 10, compiled Aug 12 2013 00:27:34) Modified by pkg...@li... Compiled by buildd@ Huge version without GUI. Features included (+) or not (-): +arabic +file_in_path +mouse_sgr +tag_binary +autocmd +find_in_path -mouse_sysmouse +tag_old_static -balloon_eval +float +mouse_urxvt -tag_any_white -browse +folding +mouse_xterm -tcl ++builtin_terms -footer +multi_byte +terminfo +byte_offset +fork() +multi_lang +termresponse +cindent +gettext -mzscheme +textobjects -clientserver -hangul_input +netbeans_intg +title -clipboard +iconv +path_extra -toolbar +cmdline_compl +insert_expand -perl +user_commands +cmdline_hist +jumplist +persistent_undo +vertsplit +cmdline_info +keymap +postscript +virtualedit +comments +langmap +printer +visual +conceal +libcall +profile +visualextra +cryptv +linebreak +python +viminfo +cscope +lispindent -python3 +vreplace +cursorbind +listcmds +quickfix +wildignore +cursorshape +localmap +reltime +wildmenu +dialog_con -lua +rightleft +windows +diff +menu -ruby +writebackup +digraphs +mksession +scrollbind -X11 -dnd +modify_fname +signs -xfontset -ebcdic +mouse +smartindent -xim +emacs_tags -mouseshape -sniff -xsmp +eval +mouse_dec +startuptime -xterm_clipboard +ex_extra +mouse_gpm +statusline -xterm_save +extra_search -mouse_jsbterm -sun_workshop +farsi +mouse_netterm +syntax system vimrc file: "$VIM/vimrc" user vimrc file: "$HOME/.vimrc" 2nd user vimrc file: "~/.vim/vimrc" user exrc file: "$HOME/.exrc" fall-back for $VIM: "/usr/share/vim" Compilation: gcc -c -I. -Iproto -DHAVE_CONFIG_H -g -O2 -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=1 Linking: gcc -Wl,-Bsymbolic-functions -Wl,-z,relro -Wl,--as-needed -o vim -lm -ltinfo -lnsl -lselinux -lacl -lattr -lgpm -ldl -L/usr/lib/python2.7/config-i386-linux-gnu -lpython2.7 -lpthread -ldl -lutil -lm -Xlinker -export-dynamic -Wl,-O1 -Wl,-Bsymbolic-functions --- Araki Ken ara...@us... |
From: Araki K. <ara...@us...> - 2014-01-17 17:15:05
|
Hi, From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Tue, 14 Jan 2014 09:13:17 +0530 Message-ID: <201...@SD...> > How does a tool like pango-view or firefox render devanagari correctly? Is > it possible to adopt what they do? I think it uses pango to render devanagari. I'm sorry but I haven't figured out how and where pango renders devanagari yet. Mlterm doesn't need full pango library but needs the conversion table from unicode codepoints to font indexes. > Would also like to understand the application's role in rendering. E.g. > if I cat a file on the terminal it renders alright, but same in > vi/more/less does not look the same, not in elinks either. Console applications usually assume that one unicode character should occupy one fixed-width column in the terminal. This doesn't hold good for indic scripts which need complex text layout. It is difficult (impossible?) for console applications to determine how many *columns* (not characters) should be overwritten or deleted in editing indic scripts. If you know some examples of console applications rendering indic scripts correctly on a terminal, please inform me of them. BTW, I found some mlterm bugs of rendering indic scripts. Mainly, mlterm converts unicode indic scripts to ISCII ones internally, but U+950 - U+95e weren't correctly converted to ISCII because each of these characters is equivalent to *two* ISCII characters. (Some unicode characters (e.g. U+960 - U+963 etc) which I couldn't find equivalent ISCII characters are still untouched.) I fixed these bugs, so try hg head. => https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz The difference between 3.3.2 and hg head is http://mlterm.sf.net/mlterm-fixindic.png Regars, PS: If you find text files which pango-view and mlterm show differently, please send it to me. --- Araki Ken ara...@us... |
From: Araki K. <ara...@us...> - 2014-01-20 16:41:05
|
Hi, From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Sat, 18 Jan 2014 13:53:17 +0530 Message-ID: <201...@SD...> > On Sat, Jan 18, 2014 at 02:14:56AM +0900, Araki Ken wrote: > > I fixed these bugs, so try hg head. > > => https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz > > I get the following error when trying to compile the above on tiny core > linux. Any leads? > > libtool: link: `x_font_xft.lo' is not a valid libtool object > make[1]: *** [libtype_xft.la] Error 1 > make: *** [all] Error 2 Will you send me a log file of ./configure and make ? If you have cygwin 1.7 environment, http://mlterm.sourceforge.net/mlterm-cygwin17-20140121.zip (http://mlterm.sf.net/bin.html) is also available. This archive contains executable binaries of the latest mlterm. (Usage) Install DVYG0ntt.ttf. Get http://mlterm.sourceforge.net/mlterm-cygwin17-20140121.zip and http://mlterm.sourceforge.net/main.indic Setup as follows. $ cd / ; unzip foo/bar/mlterm-cygwin17-20140121.zip => install binaries to /bin, /etc and /lib/mlterm. $ cp foo/bar/main.indic ~/.mlterm/main $ mlterm BTW, I fixed additional rendering problems, and updated https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz Regards, --- Araki Ken ara...@us... |
From: Mayuresh <may...@ac...> - 2014-01-21 04:41:18
Attachments:
wrongones
|
On Tue, Jan 21, 2014 at 01:52:44AM +0900, Araki Ken wrote: > I fixed and updated https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz. Yes, I notice that it is improved. I could find a few characters quickly that are wrong. See attachment. Use pango-view or lexilogos to see their correct rendering. (I created them using lexilogos, saved in a file and cated that file in mlterm.) In addition to above, I noticed "." being rendered as "?" on mlterm for text that looks alright in lexilogos, though I could not simulate that again. I am still not able to get elinks render devnagari, except for very few lines on a web page. Neither lynx is rendering the characters right. I saw your screen shots of a website. Which browser was that? Can you try this site: http://maharashtratimes.indiatimes.com/newsletter.cms Mayuresh |
From: Mayuresh <may...@ac...> - 2014-01-24 11:55:43
|
On Fri, Jan 24, 2014 at 07:39:31AM +0900, Araki Ken wrote: > The screenshot of elink is as follows. Saw it. It is looking very good and largely correct. > elinks regards the width of all characters as fixed, so complete rendering > is impossible. > (Start mlterm with -m=false option to get better rendering because elinks > doesn't support character combining.) Tried that. I still see a bunch of "*" characters only, while cating files on terminal is working fine. It is not that all characters are not rendered, a few characters are appearing right and the rest "*". Any elinks setting? Mayuresh |
From: Araki K. <ara...@us...> - 2014-01-24 14:50:35
|
Hi, From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Fri, 24 Jan 2014 17:25:36 +0530 Message-ID: <201...@SD...> > Tried that. I still see a bunch of "*" characters only, while cating files > on terminal is working fine. It is not that all characters are not > rendered, a few characters are appearing right and the rest "*". Any > elinks setting? Do you specify *.UTF-8 locale as the value of LANG and LC_* environmental variables ? What happen if you specify *.UTF-8 locale explicitly as follows ? $ LANG=en_US.UTF-8 elinks http://... --- Araki Ken ara...@us... |
From: Araki K. <ara...@us...> - 2014-01-23 22:44:44
|
Hi, From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Tue, 21 Jan 2014 10:11:08 +0530 Message-ID: <201...@SD...> > I could find a few characters quickly that are wrong. See attachment. > Use pango-view or lexilogos to see their correct rendering. (I created > them using lexilogos, saved in a file and cated that file in mlterm.) Thanks. I fixed this problem with some other rendering fixes. http://mlterm.sf.net/mlterm-fixindic2.png > I am still not able to get elinks render devnagari, except for very few > lines on a web page. Neither lynx is rendering the characters right. I saw > your screen shots of a website. Which browser was that? Can you try this > site: http://maharashtratimes.indiatimes.com/newsletter.cms The screenshot of elink is as follows. elinks regards the width of all characters as fixed, so complete rendering is impossible. (Start mlterm with -m=false option to get better rendering because elinks doesn't support character combining.) http://mlterm.sf.net/mlterm-indic-elinks.png I usually use w3m which seems to render better than elinks. http://mlterm.sf.net/mlterm-indic-w3m.png Regards, --- Araki Ken ara...@us... |
From: Araki K. <ara...@us...> - 2014-01-20 16:53:00
|
Hi, > > On Sat, Jan 18, 2014 at 02:14:56AM +0900, Araki Ken wrote: > > > I fixed these bugs, so try hg head. > > > => https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz > > > > I get the following error when trying to compile the above on tiny core > > linux. Any leads? > > > > libtool: link: `x_font_xft.lo' is not a valid libtool object > > make[1]: *** [libtype_xft.la] Error 1 > > make: *** [all] Error 2 > > Will you send me a log file of ./configure and make ? Sorry, I didn't notice my careless mistake. I fixed and updated https://bitbucket.org/arakiken/mlterm/get/tip.tar.gz. Regards, --- Araki Ken ara...@us... |
From: Araki K. <ara...@us...> - 2014-01-13 22:50:02
|
Hi, Thanks for your obvervations. From: Mayuresh <may...@ac...> Subject: Re: [Mlterm-dev-en] mlterm config Date: Mon, 13 Jan 2014 08:55:53 +0530 Message-ID: <201...@SD...> > 3. There are more popular fonts such as Lohit Marathi, which strangely do > not work. I wonder whether the space in their name is a problem. Lohit Marathi font doesn't have glyphs to render ISCII text by libind. > 7. Lastly, the rendering is very approximate and has several limitations. > It is better than some of the other terminals but inferior to what a web > browser, office or even what a tool like pango-view shows. The support of indic scripts depends on libind library which was developed 15 years ago. libind converts ISCII text to the glyph indexes of such fonts as DVYG0ntt.ttf and DVSR0ntt.ttf which contain glyphs in a different order from modern unicode fonts. * The glyphs in DVG0ntt.ttf is http://mlterm.sf.net/ttyogesh-glyphs.png * The conversion table from ISCII hindi text to the glyph indexes is https://bitbucket.org/arakiken/mlterm/src/tip/libind/table/hindi.table I'm not familiar with indic scripting, but I think the current processing depending on libind is obsolete. I would like to change it to the modern and light-weight alternative if available. I'll be happy if you report strange behaviors or bugs of libind conversion tables etc in detail or suggest alternative libraries. Regards, --- Araki Ken ara...@us... |
From: François P. <fra...@mi...> - 2014-01-14 10:12:11
Attachments:
signature.asc
|
Le 13/01/2014 23:49, Araki Ken a écrit : > Hi, > > Thanks for your obvervations. > > From: Mayuresh <may...@ac...> > Subject: Re: [Mlterm-dev-en] mlterm config > Date: Mon, 13 Jan 2014 08:55:53 +0530 > Message-ID: <201...@SD...> > >> 3. There are more popular fonts such as Lohit Marathi, which strangely do >> not work. I wonder whether the space in their name is a problem. > > Lohit Marathi font doesn't have glyphs to render ISCII text by libind. > >> 7. Lastly, the rendering is very approximate and has several limitations. >> It is better than some of the other terminals but inferior to what a web >> browser, office or even what a tool like pango-view shows. > > The support of indic scripts depends on libind library which was developed > 15 years ago. > > libind converts ISCII text to the glyph indexes of such fonts as DVYG0ntt.ttf > and DVSR0ntt.ttf which contain glyphs in a different order from modern unicode > fonts. As far as I understand, the use of mlterm with indic scripts, needs to enter iscii code in mlterm which will return glyphs from an appropriate font; am I right? So the problem is: how to enter iscii code using ibus and m17n libraries? I did not find any iscii input method in iBus... And the use of unicode fonts like FreeSerif is (up to now) impossible. Regards -- François Patte UFR de mathématiques et informatique Laboratoire CNRS MAP5, UMR 8145 Université Paris Descartes 45, rue des Saints Pères F-75270 Paris Cedex 06 Tél. +33 (0)1 8394 5849 http://www.math-info.univ-paris5.fr/~patte |
From: Mayuresh <may...@ac...> - 2014-01-14 14:57:06
|
On Tue, Jan 14, 2014 at 11:10:09AM +0100, Fran?ois Patte wrote: > As far as I understand, the use of mlterm with indic scripts, needs to > enter iscii code in mlterm which will return glyphs from an appropriate > font; am I right? Input method is another dimension altogether. First problem that appears to me is of correct rendering. BTW, as input method I prefer online itrans method: http://www.lexilogos.com/keyboard/devanagari.htm I have never got to work scim or anything like that work (on NetBSD that is). Mayuresh |