[Freeciv-Dev] Re: headers in po-files

[Gaute B Strokkenes <gs234@xxxxxxxxx>]

On Tue, 26 Jun 2001, egbert@xxxxxxxxxx wrote:
> On Tue, 26 Jun 2001, Thue wrote:
> 
> IMHO: No!
> Exception: A patch is needed because an .po file crashes the game.
> 
> Why? Changes should be done only by people really knowing what they
> are doing: I wouldn't try to change the AI Code just for fun...

What, precisely, makes you think that we don't know what we're doing?
If I wasn't sure that I knew what I was doing, then I wouldn't do it.

> Let's speak about the last patches:
> 
> (1) Do you really know which ISO every translator uses?
>     I *don't* know. Maybe you know this about all of the CVS' pos,
>     but are all people of a team using the same? I don't think so:
>     At least in "EG-Europe" a lot of people are changing between
>     ISO-8859-1 and ISO-8859-15.

You're confused.  Firstly, the patch didn't change the actual semantic
value of any of the charset declarations[1], it only changed it to a
canonical synonym.  The reason why this is necessary is that different
systems vary greatly in how liberal they are in accepting synonyms for
charset names.  For instance, on one system "latin1" might be accepted
as a synonym for "ISO-8859-1" but on another system it might not.
Therefore, the gettext manual specifies a canonical form that should
be used.  Secondly, not every charset is an "ISO".  (ISO is the
International Standardisation Organisation, so I assume that your use
of this acronym is intended to read "charset specified by an ISO
standard.)

Frankly, it would appear that in this case it is you who do not know
what you're doing, not the other way round.

> (2) Do you know about the plural forms of every language?
>     And how to define them?

I looked in the gettext manual.  In the one case where it wasn't given
(pt and pt_BR) I added a comment saying that it should be added
instead.

> I *don't* know.

This does not prevent someone else from knowing, you know.

> And: It's an *optional* (gettext manual), very new feature. Noone
> knows if it may cause any trouble. Using it, is a risk - and freeciv
> does *not* support it now nor will it do so in the near
> future. (Before (if) this ability reaches the translators, the
> programmers have to do a lot of work: Every phrase containing a
> number has to be changed...)

I'm sure it will warm your heart that some programmers, such as me, do
care about this sort of thing and are planning to look into this in
the near future.  (And no, you don't have to change every phrase that
contains a number, just some of them.)

> (3) Why change the "fuzzy" header flag?
>     As I said here, it makes no difference to use it or
>     not. S.o. didn't believed it and asked the main author of
>     gettext. He said the same.

There definitely is: with the charset information available, gettext
is able to convert the translations from whatever format the po file
is into whatever LC_CTYPE is.  For instance, this enables me to make
use the Norwegian translation in a no_NO.UTF-8 locale even though the
po file uses ISO-8859-1 and so on.  Apart from the obvious usefulness
of this, there are a lot of people, myself included, who consider
UTF-8 convergence to be a Good Thing(tm).

I find it hard to believe that Bruno told you that this change has no
effect; such a statement is in direct contradiction with both the
manual and the actual code.  I think it is rather more likely that you
have misunderstood him.

> So there is no reason to change it. But we know, not using this flag
> caused problems in the past. They maybe solved or not: There is *no*
> advantage in changing and risk in doing so.  I ask again: Why change
> it?

You may have noticed that we didn't simply remove the flags:
considerable effort was expended to find and fix the bug that caused
it not to work.  Now that we believe that the bug has been fixed, the
sensible thing to do is to change it back.  Off course, if a bug
surfaces then we will look into it again.

> Three "Non-translator"-patches. None of them really needed.

Pft.  Sure they are.

> All of them risky.

All patches of any sort whatsoever are risky.  There is always the
risk that you have not properly thought through all the consequences
whenever you make any change of any kind.  It is silly to allow this
possibility to stop you from making any change of any sort.

The only concrete risk that I can think of would be if one of plural
form definitions given in the gettext manual are wrong or were
inadvertently pasted wrongly into the files or something.  It would be
downright silly to declare that no one should make any patches because
of the remote possibility that a cut-and-past error might possibly
have occurred somewhere.

Even if there were some sort of problem with the plural form
definitions that caused Freeciv to fail catastrophically, then I'm
pretty sure that someone would make a note of the fact and inform
freeciv-dev, at which point some sort of action would be taken.  This
is not the sort of cataclysmic and irreversible doomsday scenario that
you appear to make it out to be.

> They *are* reasons to say .po files are handled by their local
> maintainers only. IMHO: good reasons.

It's easy to see why programmers should keep their hands of the
translations proper.  I don't see how this carries over to the sort of
minor technical detail that you are complaining about.  Please feel
free to list all the reasons that you have in mind.

Frankly, I think that your portrayal of the people who come up with
patches and the people who approve them as careless brutes who
randomly clobber things they do not really understand is
ill-considered and rather rude, though I'm sure that it's not intended
to come across that way.  We do try hard to understand what we are
changing whenever we change something, and in this case it seems that
my understanding of the issues involved is better than yours and that
I've done my homework better than you have.  I also rather resent the
implication that these changes were carried out behind the collective
backs of the translators: the patches were posted to freeciv-dev for
public review, just like any other patch.  For the record, I'm all for
creating a separate forum where such patches could be posted (in
addition to freeciv-dev, off course) so that translators don't have to
be burdened with reading all of freeciv-dev.

I have no problem understanding that it can be very annoying to have
someone else appear and make changes to "your" stuff.  This happens to
everyone; sometimes you post a patch and you're told to change it in a
way that makes it more palatable for the people in charge and
sometimes you have the change made for you, frequently without being
asked first.  Sometimes this is annoying because you think that the
original was better anyway.  Dealing with stuff like this is
completely necessary if you want to contribute to a free software
project, because it's guaranteed to happen sooner or later.

[1] With the exception of sv where the appropriate declaration was
missing.  However, since I stem from that corner of the world I know
perfectly well what charset sv.po uses.

-- 
Big Gaute                               http://www.srcf.ucam.org/~gs234/
Hmmm..  A hash-singer and a cross-eyed guy were SLEEPING on a deserted
 island, when...