Re: gettext improvements (was: lynx-dev French messages file)

[Webmaster Jim]

On Tue, Jun 08, 1999 at 11:53:12AM +0900, Henry Nelson wrote:
> > Also, Japanese bz2 file could not be read properly since SLCC server
> Jim, if this is my file, I would prefer that it remain private.  Link
...
> cases purposely blanks out messages. (It is by no means complete.) You
> would be well advised to use the one distributed with Hiroyuki Senshu's
> Win32 package if you need a ja.po file.

All right; I will extract that and put it on SLCC.

> 
> > > There are still many messages to be created, but over 1000 are working.
> There are also many messages which could be done away with by combining
> with others and reusing them more efficiently.  (Slowly, I will submit
> patches.)  One example right off the top of my head is this pair in
> LYMessages_en.h, where what is important is that the field cannot be
> changed, not what kind of field it is:
> #define FORM_LINK_TEXT_UNM_MSG \
>  gettext("UNMODIFIABLE form text field.  Use UP ...
> #define FORM_LINK_TEXT_SUBMIT_UNM_MSG \
>  gettext("UNMODIFIABLE form field.  Use UP ...

Yes. There are other places I noticed where the message is prefixed with
the module name. That does not belong in the translated text. However,
I've been spending my time begging for translation files rather than
tweaking the existing code.


> Also, I hope putting all of the .lynxrc comments within gettext() is not
> going to set a trend because that will create a huge catalogue of messages,
> making the translators task formidable (for me, impossible).  The information
> is readily available in other files, not even considering whether or not we
> want users to be hand editing .lynxrc.  In general, it is easier to translate
> text/html files in their entirety than to translate long, multi-line gettext
> strings.  One example of what I mean given below.  This:
> 
> src/LYrcFile.c      [ write_list(fp, gettext("\ ... ]
> default selection.  When this option is set to \"advanced\", and the\n\
> user mode is advanced, the 'v'iew bookmark command will invoke a\n\
> statusline prompt instead of the menu seen in novice and intermediate\n\
> user modes.  When this option is set to \"standard\", the menu will be\n\
> 
> is better done (and IMO more easily translated) here:
> 
> lynx_help/keystrokes/option_help.html     [as rendered in Lynx]
>      * If set to STANDARD, a menu of available bookmarks is always
>        invoked when you seek to view a bookmark file or add a link, and
>        you select the bookmark file by its letter token in that menu.
>      * If set to ADVANCED, you are instead prompted for the letter of the
>        desired bookmark file, but can enter '=' to invoke the STANDARD
>        selection menu, or RETURN for the default bookmark file.
> 
> and is also available in different wording here:
> 
> distribution lynx.cfg:      [ available as a link off of =)Info page ]
> # default via the .lynxrc file.  When on, the setting can be STANDARD or
> # ADVANCED.  If SUPPORT is set to the latter, and the user mode also is
> # ADVANCED, the VIEW_BOOKMARK command will invoke a statusline prompt at
> # which the user can enter the letter token (A - Z) of the desired bookmark,
> # or '=' to get a menu of available bookmark files.  The menu always is

Right. The long messages in general are losers, because as soon as
someone "improves" the grammar the entire translated message is out of
synch and thus lost. I see this in the French messages. Perhaps the
effort to simplify the message base can be combined with the effort to
simplify the options control flow.

++++++++++++++++++++++++++++
Marvin the Paranoid Android.