Federico Lucifredi wrote:
(please CC: at least lfs-dev)
Hello Jim, Alex,
Yes, I have plans for (1) and (2), so there should be no particular
problem getting that fixed. I have not thought extensively of the
interaction problems with groff, tho, so that is next on the list. for
(3), I would like to know exactly what behavior you would like to see --
install time switch, commandline at invokation time, or what else.
3) Feature request: it would be very nice if LFS obtains a way to tell
Man to ignore /usr/share/man/ja/* even in Japanese locales, because the
system's Groff-1.19.{2,3cvs} can't format those manuals. This also
applies to other languages, and maybe it is better to implement as a
whitelist, not blacklist. This whitelist should be different for
printing and display purposes.
I want a new option in man.conf. Basically, it's a way for root of a box
(which, e.g., has many Russian users but allows one German person to ssh
in and override $LANG), to say: "I have created a man setup that works
only for Russian, don't attempt to use it in other situations". The idea
is to never misformat a manual page unless the "yes --help" output would
also be misformatted. Fallback to English is much better than misformatting.
This approach is different from the one of Man-DB where it knows good
defaults for many locales (but still misformats Chinese manuals in
zh_CN.UTF-8).
Let's name the option "TRANSLATIONS" for now. Its value should be a
colon-separated list of the following allowable items:
1) language names, such as de, fr,..., that contain only letters.
2) the special tokens "8bit" and "utf8"
3) exact locale names, such as ja_JP. They can be distinguished from (1)
by the presence of non-letters.
Similarly, HARDCOPY_TRANSLATIONS can list the same items.
The distributor or whoever else creates man.conf should set these
options to the list of languages/locales where manual pages are known to
be formatted properly by *roff and other programs mentioned in man.conf.
See below for the exact meaning.
Currently, Man looks at locale environment variables and the LANGUAGE
variable to determine the list of search paths for translated manual
pages, and picks up the first manual page that exists in the search
path. Proposed change: ignore bad manual pages, according to the
following rules:
1) if MB_CUR_MAX==1 in the current locale, and the special "8bit" token
is precent in TRANSLATIONS, any manual page is good. The use case is the
current -Tlatin1 setup which formats all manuals properly in 8-bit
locales if they are stored in 8-bit language-specific encodings.
2) if the currenl locale is a UTF-8 based locale, and the "utf8" token
is present in TRANSLATIONS, any manual page is good. The use case is
RedHat's groff.
3) if the exact current locale is listed in TRANSLATIONS, and the manual
page doesn't come from the LANGUAGE environment variable, it is good.
The use case is: format Japanese manuals with -Tnippon on Debian-patched
groff.
4) if the manual page's language is listed as a language (not locale!)
in TRANSLATIONS, it is good. Use case 1: no -Txxxx switch, rely upon the
nroff script itself to figure out the correct device for
ISO-8859-1-encoded manuals. Use case 2: the -K switch for the new groff
(because the argument correct for Russian manual pages is wrong for
German ones).
5) English manual pages are always good.
BTW, I would like to avoid (4.2) and the whole need for the
administrator to hardcode the right -K ... argument, because this can be
deduced from the manual page location. The old idea about
/usr/share/man/$lang/.charset file may help, maybe it would be a good
idea to introduce the "%C" token that would expand to that, and "%K"
that would expand to "-K %C" if "%C" expands to non-empty string. The
default value for "%C" for manual pages specified with the full path is
a tough question, I will think more about it. Maybe: empty string.
Any other manual page is bad and should be treated as a non-existing file.
The good default would be (assuming the Debian-specific manual page
encoding, -Tlatin1 default nroff argument, and the "less -isR" pager
that is able to convert from ISO-8859-1 to UTF-8 on the fly):
TRANSLATIONS 8bit:da:de:en:es:fi:fr:ga:gl:id:is:it:nb:nl:nn:no:pt:sv
HARDCOPY_TRANSLATIONS da:de:en:es:fi:fr:ga:gl:id:is:it:nb:nl:nn:no:pt:sv
--
Alexander E. Patrakov
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
