Re: lynx-dev NLS

[Henry Nelson]

> > If you are going to create two distributions, then create one _complete_
> > NLS distribution, and another distribution _entirely devoid_ of the
> > gettext *package*, i.e., no intl/ and no po/ subdirectories, and no
> > ABOUT-NLS file.  
> 
> That was my point and the reason I used the word devoid - which

I wasn't sure; I thought you were concerned only with the po files.  My
proposal is for one distribution with the gettext package (/intl and
ABOUT-NLS) and all the [almost] complete po files (/po), and another
distribution which has NONE of the nls stuff.  I think it would only
make sense to do this at each release.  I think for the development
versions, the present "stub file" system is adequate since Tom needs
it there handy for testing.  However, the "stubs" certainly ought to
be representations of real files somewhere.  The last I looked, the
majority were stubs of fictitious files.  I definitely agree that the
"stub file" system should not carry over to any "official" release bundles.

In the past I have also proposed splitting the distribution into two:
the core code, and the docs and misc.  It is not as if this is a new
idea; it has been done to a number of major applications out there.
I think doing this would make it more practical to include the complete
translation catalogues for all languages they are available for.  One
big consideration is that a complete lynx.po file runs over 150kB.

> I just don't see the usefulness of including half finished work
> anywhere - then rather do without it, but that's truly a matter
> of opinion.

I tend to think if the file has a third or more of the strings translated,
then it ought to be included until a more complete one is provided.  But,
I agree it is a matter of opinion, and I could go either way.  If you are
talking about all the 2-3kB po files that were in the distribution originally,
then yes, they should be removed _immediately_ if they are still in there.

> But I really can't get any closer to what I think needs to be
> done than in my previous posting. Your remarks, while
> commendable, don't change anything.

I know your feeling.  It's frustrating to say the least.  I've been
talking to the wall since the very inception of gettext in lynx.

If you have the time and interest, however, your proposals might carry
more weight if you were to list up the "complete and correct," the
"partial," and the "should be deleted" po files.  What are the po files
that are actively maintained: da, ja, ru, de?  Which ones need work: fr?
Which ones never should have been there in the first place: ko, it, pt?
Are there any that should be in the po directory, but aren't?  State
your criteria.  If no one acts on a concrete proposal to preserve the
complete files, tag or move to a sub-directory of po/ the incomplete files
and delete the fictitious files, then we may as well pack up our bags.

__Henry

; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden