Introducing open-tran.eu
Jacek Śliwerski
http://open-tran.eu. It lets people search English phrases used in open
source software for their translation into 87 languages. Right now one
can search for translations of KDE and Mozilla and so I thought that you
might be interested in it.
I will be grateful for any feedback and criticism.
--
Jacek
Axel Hecht
Would you mind giving some additional background? Like, AFAICT, you're
using existing translations of KDE and Mozilla and search ... now I'm
stuck. And where do you get the existing translations from? It'd be
interesting to learn if those are only released ones, etc.
Axel
Jacek Śliwerski
>
> Would you mind giving some additional background? Like, AFAICT, you're
> using existing translations of KDE and Mozilla and search ... now I'm
> stuck. And where do you get the existing translations from? It'd be
> interesting to learn if those are only released ones, etc.
I have downloaded the Mozilla translations from the following repository:
cvs-mirror.mozilla.org:/l10n
using the branch MOZILLA_1_8_BRANCH.
KDE translations are downloaded in a very similar manner from the stable
branch of the anonymous subversion server of the project.
Some information about it is available at http://open-tran.eu/db.html.
Is there anything I miss due to my choice of the repository and/or branch?
--
Jacek
Toni Hermoso Pulido
> Few weeks ago I have launched a service for open source translators at
> http://open-tran.eu. It lets people search English phrases used in open
> source software for their translation into 87 languages. Right now one
> can search for translations of KDE and Mozilla and so I thought that you
> might be interested in it.
>
> I will be grateful for any feedback and criticism.
Cool. Having direct links to the PO files where suggestions have been
extracted would be very nice :)
Simos Xenitellis
> Few weeks ago I have launched a service for open source translators at
> http://open-tran.eu. It lets people search English phrases used in open
> source software for their translation into 87 languages. Right now one
> can search for translations of KDE and Mozilla and so I thought that you
> might be interested in it.
>
> I will be grateful for any feedback and criticism.
Great work Jacek!
If you would like to add in the GNOME translations, I believe you can
grab them with
wget -N -A el.po -r http://l10n.gnome.org/languages/el/gnome-2-18
The above example is for the Greek "el" translations. The current stable
version of GNOME is 2.18. You can find all GNOME releases at
http://l10n.gnome.org/releases/
All the best,
Simos
Axel Hecht
You got some languages that are not released, and thus are not
necessarily in a good shape.
The list of currently shipped languages is at
http://mxr.mozilla.org/mozilla1.8/source/browser/locales/shipped-locales.
The branch name is MOZILLA_1_8_BRANCH, really, if I may nitpick on this.
I'm still not really sure as to what happens when you search, could you
elaborate on that?
Axel
Jacek Śliwerski
>
> If you would like to add in the GNOME translations, I believe you can
> grab them with
While you were sending your email, Gnome translations were being
imported to the database ;]
It seems to me that open-tran.eu is the world's largest technical
dictionary, but if you know any larger - let me know.
Anyways, I'd like to thank you all for your contribution to the Mozilla
project. Open-Tran.eu wouldn't exist without your hard work.
--
Jacek
Jacek Śliwerski
>
> You got some languages that are not released, and thus are not
> necessarily in a good shape.
There are 6 languages imported to open-tran's database that are not
currently shipped. These are: Romanian, Albanian, Swati, Veneda, Xhoan
and Zulu. Currently I have no reason to exclude them, but if people
start complaining, I will reconsider my decision.
> The branch name is MOZILLA_1_8_BRANCH, really, if I may nitpick on this.
Corrected.
> I'm still not really sure as to what happens when you search, could you
> elaborate on that?
Sure. Let us assume that we are searching for suggestions for the
phrase S. For each phrase P in the database we compute the number of
words in the symmetric difference between P and S. The lower, the
better. Notice, that by the definition of this algorithm, open-tran.eu
is guaranteed to return the exact match as the first one if such exists.
This project is open source as well and you can download its source code
from the subversion repository:
http://code.google.com/p/open-tran/source
Best regards,
--
Jacek
PS. For those who are afraid of Google - open-tran.eu is not affiliated
by this company.
Djihed Afifi
It would be superb if good robust algorithms close to human accuracy
can be implemented. I realise that's a per language problem and that
it is quite hard.
Thanks for the hard work.
Djihed
> _______________________________________________
> dev-l10n mailing list
> dev-...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-l10n
>
Julen
> Few weeks ago I have launched a service for open source translators at
> http://open-tran.eu. It lets people search English phrases used in open
> source software for their translation into 87 languages. Right now one
> can search for translations of KDE and Mozilla and so I thought that you
> might be interested in it.
>
> I will be grateful for any feedback and criticism.
> Jacek
First of all, I think you've done a great job so thank you very much
for sharing this tool with all the comunity.
Anyway, I would like to point out different aspects with the aim of
being constructive:
- It would be very fancy to have this online data-base integrated with
other online tools like Pootle, in addition to its own terminology
data-base.
- When you upload a file to translate, the results page can grow a
lot. To prevent this, a simple page pagination can fix this behavior.
- Some entries could be omitted in the search, specially those which
have variables. For example:
<b>Immediate Downloads:</b>
1. <b>%1</b> - <b>%2</b> [<b>%3</b>] (KDE amarok)
- When the searched strings contain variables such as %s, the results
are very confusing and they are far from the expected ones. Example:
Denied file request: %s
1. %1$S %2$S: %3$S %4$S (Mozilla mail ldapAutoCompErrs)
- It would be nice to be able to see the context in each entry.
Hoping this ideas can be helpful,
Julen.
Jacek Śliwerski
>
> - It would be very fancy to have this online data-base integrated with
> other online tools like Pootle, in addition to its own terminology
> data-base.
I know that the maintainers of the WordForge project are reading this
mailing list and I hope we will be able to find a solution.
> - When you upload a file to translate, the results page can grow a
> lot. To prevent this, a simple page pagination can fix this behavior.
I'm not really fond of this functionality. As long as you cannot use
open-tran to take advantage of the suggestions to really edit and save
the file, it seems useless to me - nobody will copy the suggestions from
the browser to the localization tool. I was thinking about supplying
checkboxes - so that you could choose a variant you like and then
download the file back with suggestions... but am skeptical about it. I
will take a look at Pootle and see, whether it wouldn't be better to
create a synergy of those two instead ;]
> - Some entries could be omitted in the search, specially those which
> have variables. For example:
The first phase of the project is almost over. The database has been
populated with a huge amount of data and it is a high time to go further
on and clean it up a little bit. I keep your suggestions in mind.
> - It would be nice to be able to see the context in each entry.
The design of the site needs restructuring in order to accommodate
additional information about translations.
Thank you very much for your feedback.
--
Jacek
Axel Hecht
I can't search for "speichern unter" on the german part, though, so I
take it that you only search for English phrases in the English phrases?
Axel
Axel Hecht
One thing that just came up to my mind, you need to explain how you're
dealing with licenses.
As of now, I expect the original licenses to stick to the data, which is
likely causing problems when trying to use suggestions for Mozilla
localizations, which need triple license.
Axel
Jacek Śliwerski
>
> I can't search for "speichern unter" on the german part, though, so I
> take it that you only search for English phrases in the English phrases?
You are right. Open-Tran lets you only search translation of English
phrases. This is due to the fact, that the complexity of my approach
depends linearly on the number of phrases in the source language. So I
had two choices:
1. either import a limited number of languages and let people translate
from whatever language they want to any other,
2. or import every language I could, but limiting the source language to
English only.
I have opted for the latter and do not regret it. Top 10 languages
account for about half of the queries.
Best regards,
--
Jacek
Jacek Śliwerski
>
> One thing that just came up to my mind, you need to explain how you're
> dealing with licenses.
I don't ;] I have sent my email to your mailing list and thus
acknowledged you that your work has been used in the open-tran.eu
project (which is open-source itself). I have sent similar e-mails to
KDE and Gnome lists and credited your work at
http://open-tran.eu/db.html and project's blog
(http://open-tran.blogspot.com).
If someone complains about using her or his work in open-tran.eu, I will
start reading the licenses. Right now, I don't see any reason to worry
about it in advance.
Best regards,
--
Jacek
Gervase Markham
> Axel Hecht wrote:
>> One thing that just came up to my mind, you need to explain how you're
>> dealing with licenses.
>
> I don't ;] I have sent my email to your mailing list and thus
> acknowledged you that your work has been used in the open-tran.eu
> project (which is open-source itself). I have sent similar e-mails to
> KDE and Gnome lists and credited your work at
> http://open-tran.eu/db.html and project's blog
> (http://open-tran.blogspot.com).
"Open source" is not the same thing as "acknowledge use of work" or
"must give credit". Licenses can and often do impose more requirements
than that.
> If someone complains about using her or his work in open-tran.eu, I will
> start reading the licenses. Right now, I don't see any reason to worry
> about it in advance.
Are you sure you want to take that attitude? I promise you, you don't
want the Mozilla Foundation to have to make a complaint of copyright
infringement against you.
I would suggest the following. The code for all of the projects you
import is available under the GPL (among other things, in the case of
Mozilla). People can know they are using the output of your service
legally as long as you acknowledge this, and tell people that
translations obtained in this way can only be used in projects licensed
under the GPL only.
That doesn't include Mozilla, sadly, but as long as you follow the
licences of our code, you can use it for anything you like. We don't
mind you using it to make other people's lives easier.
If people use translations provided by your service in other, non-GPLed
projects, they may be committing copyright infringement.
Gerv
Damjan Georgievski
> "must give credit". Licenses can and often do impose more requirements
> than that.
>
>> If someone complains about using her or his work in open-tran.eu, I will
>> start reading the licenses. Right now, I don't see any reason to worry
>> about it in advance.
>
> Are you sure you want to take that attitude? I promise you, you don't
> want the Mozilla Foundation to have to make a complaint of copyright
> infringement against you.
Are translations even copyright-able?!?
--
damjan
Axel Hecht
We consider translations to be modifications of the original work, and
as such, they carry at least one of the original licenses. Localizations
in CVS are actually expected to be triple licensed.
Axel
Jacek Śliwerski
>
> "Open source" is not the same thing as "acknowledge use of work" or
> "must give credit". Licenses can and often do impose more requirements
> than that.
Sure they can. I am aware of that. I wanted to stress that I act in a
good will and gave you an option to react. If you want me to remove
Mozilla's translations from my database - please express it explicitly
and I will do so.
> Are you sure you want to take that attitude? I promise you, you don't
> want the Mozilla Foundation to have to make a complaint of copyright
> infringement against you.
Yes, I am sure. In fact - Mozilla's copyright infringement against
open-tran.eu could bring hundreds of thousands of people to my website.
Filing a complaint against me, in the current situation, sounds
barratrous to me and I am not afraid of that.
> If people use translations provided by your service in other, non-GPLed
> projects, they may be committing copyright infringement.
Do you mean that translating "Save as..." into Polish as "Zapisz
jako..." (providing that you found it at open-tran.eu) is a copyright
infringement against Mozilla?
So how open-tran.eu violates the license terms of Mozilla's translations?
Best regards,
--
Jacek
Toni Hermoso Pulido
between people who do not know each other, I think Gervase only wanted
to warn you about the implications of the licenses.
And, sadly or not, I also agree you should consider them, not because
Mozilla Foundation might "threat" you, which I would ever think it would...
As far as I can say, I think that just adding a noticeable warning such
as Simos suggested, would be sufficient. Thus, it's simply the
responsibility of the user, not yours, what they do with the strings.
As I told in a previous message, I would encourage you to add a link to
the very files in the suggestion list; this would help to track them,
also regarding to the licenses.
Cheers,
En/na Jacek Śliwerski ha escrit:
> Gervase Markham wrote:
>> "Open source" is not the same thing as "acknowledge use of work" or
>> "must give credit". Licenses can and often do impose more requirements
>> than that.
>
> Sure they can. I am aware of that. I wanted to stress that I act in a
> good will and gave you an option to react. If you want me to remove
> Mozilla's translations from my database - please express it explicitly
> and I will do so.
>
>> Are you sure you want to take that attitude? I promise you, you don't
>> want the Mozilla Foundation to have to make a complaint of copyright
>> infringement against you.
>
> Yes, I am sure. In fact - Mozilla's copyright infringement against
> open-tran.eu could bring hundreds of thousands of people to my website.
>
> Filing a complaint against me, in the current situation, sounds
> barratrous to me and I am not afraid of that.
>
>> If people use translations provided by your service in other, non-GPLed
>> projects, they may be committing copyright infringement.
>
Ville Pohjanheimo
Uh IANAL, but... I think various countries/languages would have "prior
art" sort of exceptions for the words used in localizations (ie. File,
Edit, About etc.) even if single word translations could be copyrighted.
However, each file separately as a database of sort, having a syntax
etc. surely must have copyright and license protection.
So (e.g.) the Finnish translation of Firefox is copyrighted and
licensed, but the translation of "File" in that set of translations most
certainly is not.
How that translates to a dictionary of localized strings ... is not at
all clear to me though I think all and every kind of proverbial sabre
rattling here is just silly (no, I'm not referring to you Axel).
-ville/mozilla.fi
Axel Hecht
Well, "tainting" strings in a particular way is hard, and I don't think
this is exact science. Note, AFAICT, open-tran lives from the fact that
it is some kind of mash-up of different projects, and those need to pay
attention to the license that particular content is under. I'd be
surprised if the GPL would be fine with licensed content just giving
such a formulation.
Axel
Axel Hecht
> Before this might get hotter, as sometimes happen with email messages
> between people who do not know each other, I think Gervase only wanted
> to warn you about the implications of the licenses.
> And, sadly or not, I also agree you should consider them, not because
> Mozilla Foundation might "threat" you, which I would ever think it would...
> As far as I can say, I think that just adding a noticeable warning such
> as Simos suggested, would be sufficient. Thus, it's simply the
> responsibility of the user, not yours, what they do with the strings.
> As I told in a previous message, I would encourage you to add a link to
> the very files in the suggestion list; this would help to track them,
> also regarding to the licenses.
Due to bad tools, most l10n files in our CVS repository don't come with
license headers themselves, and those that do don't necessarily come
with the appropriate ones. Like, find me an example in our tree where
the license plate shows, has the right license, and contributor list. Fun.
Axel
Axel Hecht
We didn't say that, neither Gerv nor me.
What we said is that we're assuming you're using our code under the
provision of the GPL/LGPL, the KDE strings are likely a mix of that, too.
That means nothing less than, Mozilla localizers can't use your site.
That doesn't mean that you can't use our code. The benefit of our triple
license, nothing ever gets back upstream.
Axel
Damjan Georgievski
>> want the Mozilla Foundation to have to make a complaint of copyright
>> infringement against you.
>
> Yes, I am sure. In fact - Mozilla's copyright infringement against
> open-tran.eu could bring hundreds of thousands of people to my website.
>
> Filing a complaint against me, in the current situation, sounds
> barratrous to me and I am not afraid of that.
And if that happens I'll make sure you get the macedonian translation under
very liberal license... Why not, even Public Domain.
--
damjan
Damjan Georgievski
> Due to bad tools,
You are poiting out Mozilla tools flaws - Cancel or Allow?
And here am I thinking it's Cancel by default.
--
damjan
Robert Kaiser
> Gervase Markham wrote:
>> If people use translations provided by your service in other,
>> non-GPLed projects, they may be committing copyright infringement.
>
> Do you mean that translating "Save as..." into Polish as "Zapisz
> jako..." (providing that you found it at open-tran.eu) is a copyright
> infringement against Mozilla?
Actually, most likely the other way round. Taking a string from KDE or
GNOME and using it in a Mozilla L10n is most likely a copyright
infringement.
The good thing with the Mozilla tri-license is that anyone licensing his
Mozilla-based work in any way allowed by either of the three licenses
can do so.
The bad thing is, we can only use stuff licensed under _all three of
them_, not stuff licensed in only one or two of them.
So no GPL-only-licensed strings can be used by a Mozilla localizer,
probably, while a Mozilla-tri-licensed string can be happily used by a
localizer of KDE, GNOME or whatever MPL- or GPL- or LGPL-licensed project.
It's also important to not though that any of those strings can surely
not be used under a license that is not compatible with those OSS licenses.
Robert Kaiser
Simos Xenitellis
How do you recommend we should deal with the license issue?
a. Put some sort of disclaimer that places the onus on the end-user.
What's the wording you recommend.
b. Remove the Mozilla translations.
c. Ask somewhere else (where?).
Simos
Gervase Markham
> Are translations even copyright-able?!?
Of course. They are creative works of authorship. Unless, of course, you
think we can sack all of our l10n teams and replace them with a computer
program which does automatic search-and-replace? :-)
Gerv
Gervase Markham
> Well, "tainting" strings in a particular way is hard, and I don't think
> this is exact science. Note, AFAICT, open-tran lives from the fact that
> it is some kind of mash-up of different projects, and those need to pay
> attention to the license that particular content is under. I'd be
> surprised if the GPL would be fine with licensed content just giving
> such a formulation.
While "The translated text carries the distribution license of the
respective projects it comes from." would be a true statement, it would
neither be very helpful to the user of open-tran.eu, nor would it
satisfy the requirements of some of the licences (e.g. the GPL) which
requires explicit notice of rights.
Gerv
Gervase Markham
> How do you recommend we should deal with the license issue?
Note that we can make recommendations, but at the end of the day,
compliance with the law is your own responsibility.
> a. Put some sort of disclaimer that places the onus on the end-user.
> What's the wording you recommend.
I would suggest putting a link next to each translation like the following:
File 1. Fichier (KDE kdelibs (<a>GPL</a>), Mozilla editor editorOverlay
(<a>MPL/LGPL/GPL</a>),...
so the link name is a list of licences, and the link target is a page
explaining about that licence or set of licences.
You could then also offer a parameter to the service so that it only
returned correctly-licensed translations. E.g. I could say "I want only
translations compatible with the GPL", and it just wouldn't return e.g.
Apache-licensed translation suggestions.
> b. Remove the Mozilla translations.
It's not just the Mozilla translations. This issue applies to all of the
free software translations you include, unless perhaps they are from
BSD-licensed projects.
Gerv
Gervase Markham
> Uh IANAL, but... I think various countries/languages would have "prior
> art" sort of exceptions for the words used in localizations (ie. File,
> Edit, About etc.) even if single word translations could be copyrighted.
"Prior art" is a trademark term, not a copyright one.
The fact that "File" translates to "Fichier" in French is not
copyrightable. However, if I write a book which contains thousands of
English words, and someone translates it into French, the result is a
derivative work of mine even if each individual translated word isn't.
Similarly, if an e.g. MPL-only project took one "GPLed" string from
open-tran.eu, there probably wouldn't be a problem. But 10, or 100, or
1000? There's no hard limit - ask a judge. But just because there's no
hard limit doesn't mean there's no limit. 1 string is OK; 1000 is too many.
Gerv
Gervase Markham
> Sure they can. I am aware of that. I wanted to stress that I act in a
> good will and gave you an option to react. If you want me to remove
> Mozilla's translations from my database - please express it explicitly
> and I will do so.
We are happy for you to do whatever you like with the translations as
long as you respect the terms of the licence.
> Filing a complaint against me, in the current situation, sounds
> barratrous to me and I am not afraid of that.
I promise you, it's not. I would not be having this conversation with
you if I thought that what you are doing is legal. What would be the
point? We are a free software organisation. We are generally in favour
of people reusing our stuff under the terms of our licences. Sueing
legitimate users would be incredibly dumb.
>> If people use translations provided by your service in other,
>> non-GPLed projects, they may be committing copyright infringement.
>
> Do you mean that translating "Save as..." into Polish as "Zapisz
> jako..." (providing that you found it at open-tran.eu) is a copyright
> infringement against Mozilla?
No. I mean that if open-tran.eu takes the MPL/LGPL/GPL Mozilla files and
provides (from them) the translation of "Save as..." into "Zapisz
jako..." to a website user, and that person copies and pastes that
string into a BSD-licensed application, that person is beginning to
commit copyright infringement.
Now perhaps one string would not be held by a court to be committing
copyright infringement. But 1000 probably would. Where's the line? Ask a
judge. However, if I want to make a copy of a copyrighted book you have
written, it's no defence to say that I copied it one small sentence at a
time.
Copyright is not like trademark. The phrase "She went to the shops" may
be part of a hundred different copyrighted works, and that's fine. But
if you copy it out of one particular one and into your work, you are
committing copyright infringement. (For simplicity, I am ignoring the
fact that my example is a very short phrase.)
You need to make your users aware of the license attached to each string
you provide.
open-tran.eu has the potential to be a wonderfully useful service. I
would love to see it succeed. But it should not do at the expense of the
wishes (as expressed in the licence) of the copyright holders of the
material it is using.
Gerv
Axel Hecht
I don't get that question.
Axel
Simos Xenitellis
Thanks for the reply.
A quick (possibly rookie) question.
Wasn't the purpose of Mozilla to be triple-licensed so that people can
pick and choose which license to use? The fact that Mozilla is triple
licensed should not mean I have to abide to all three different licenses
at the same time.
I download the Mozilla source and I only follow the GPL license. What's
wrong with that?
Simos
> _______________________________________________
> dev-l10n mailing list
> dev-...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-l10n
Gervase Markham
> Wasn't the purpose of Mozilla to be triple-licensed so that people can
> pick and choose which license to use? The fact that Mozilla is triple
> licensed should not mean I have to abide to all three different licenses
> at the same time.
That is correct.
> I download the Mozilla source and I only follow the GPL license. What's
> wrong with that?
OK, let's say you are using only the GPL for all of the translations you
have imported (for simplicity). Then the problems are as follows:
Firstly, web pages such as this one:
http://fr.open-tran.eu/suggest/File
are derivative works of several GPLed files. Therefore, the page itself
is either under the GPL or you can't distribute it at all (GPL, section
4). However, you have not followed the provisions of GPL 2.a) or 2.b).
(There are no licensing terms attached to the page.)
The second problem is that, because of the first problem, you are
encouraging your users to violate copyright law. If someone finds this
great translation service and copies strings into a BSD-licensed app,
they are breaking copyright law, because their app will not be licensed
under GPL terms, as it is required to be if it contains GPLed code.
As I have said in other messages, these problems can be avoided if you
state clearly on your translation pages that the strings are provided
under the terms of the GPL (with a link to the license and an
explanation), and tell people that they can only be copied into GPLed
programs.
Gerv
Robert Kaiser
> 1 string is OK; 1000 is too many.
And even there, "OK" doesn't mean "it's no copyright infringement", only
"the argument is hard to hold up and probably not worth it" ;-)
Robert Kaiser
João Miguel Neves
upstream (Mozilla Firefox) then you must triple license it. That's why
Axel was noting that people can use Firefox's code/translations, but
there are more restrictions to get those changes to upstream.
For instance, a GPL-only Firefox could take advantage of all the GPL
code in existance. The triple-licensed one, can't.
Best regards,
João Miguel Neves
Sex, 2007-03-30 às 17:36 +0800, Simos Xenitellis escreveu:
> Thanks for the reply.
>
> A quick (possibly rookie) question.
>
> Wasn't the purpose of Mozilla to be triple-licensed so that people can
> pick and choose which license to use? The fact that Mozilla is triple
> licensed should not mean I have to abide to all three different licenses
> at the same time.
>
> I download the Mozilla source and I only follow the GPL license. What's
> wrong with that?
>
> Simos
>
> On Fri, 2007-03-30 at 09:57 +0100, Gervase Markham wrote:
Peter Weilbacher
> One thing that just came up to my mind, you need to explain how you're
> dealing with licenses.
>
> As of now, I expect the original licenses to stick to the data, which is
> likely causing problems when trying to use suggestions for Mozilla
> localizations, which need triple license.
Somewhat related question: how do you currently ensure that words used in
Mozilla translations are not coming from copyrighted code? If a translator
looks up a hundred (or hundreds of) words by comparing Windows versions of
English and his language and includes them into Mozilla, would that not be
a copyright problem?
Peter.
Simos Xenitellis
Thanks all for the replies.
What I see we need to write down here is what Jacek should exactly do so
that we are all happy, projects and translators. I am trying to think
how to make it comprehensive, yet usable by end users.
I would really dislike to put excessive "restrictions" on open-tran.eu
that would kill of the project.
When I browser the Mozilla translations using LXR, I do not see any
notification that the translation files are licensed, or even
triple-licensed. For example, see
http://lxr.mozilla.org/l10n-mozilla1.8/source/el/browser/chrome/browser/preferences/content.dtd
If you start from http://lxr.mozilla.org/ and you take all the way to
the translations of a language, you do not see a message regarding the
distribution license.
Is this unintentional? I suppose someone could grab the translations
from LXR and consider they are in the public domain.
In order to simplify the process, I see fit to add an extra page (or
paragraph, perhaps at the first page of open-tran.eu), titled "Licensing
issues" with content:
"This service includes the translation files of several free and
open-source projects. Apparently you may use the translation results of
a project (such as GNOME and KDE) as long as your project is licensed
with the same license. For example, both GNOME and KDE carry the GPL
license; therefore if your program is also licensed with the GPL license
you have the freedom to use this service. For a list of the licenses of
each project, see http://open-tran.eu/db.html [[Jacek adds an extra
column with distribution license]]. If you have further questions,
please consult the organisation behind your distribution license."
Would that be ok?
Simos
On Fri, 2007-03-30 at 13:17 +0100, João Miguel Neves wrote:
> You can choose only abide by the GPL. But if you want to contribute to
> upstream (Mozilla Firefox) then you must triple license it. That's why
> Axel was noting that people can use Firefox's code/translations, but
> there are more restrictions to get those changes to upstream.
>
> For instance, a GPL-only Firefox could take advantage of all the GPL
> code in existance. The triple-licensed one, can't.
>
> Best regards,
> João Miguel Neves
>
> Sex, 2007-03-30 às 17:36 +0800, Simos Xenitellis escreveu:
> > Thanks for the reply.
> >
> > A quick (possibly rookie) question.
> >
> > Wasn't the purpose of Mozilla to be triple-licensed so that people can
> > pick and choose which license to use? The fact that Mozilla is triple
> > licensed should not mean I have to abide to all three different licenses
> > at the same time.
> >
> > I download the Mozilla source and I only follow the GPL license. What's
> > wrong with that?
> >
> > Simos
> >
> > On Fri, 2007-03-30 at 09:57 +0100, Gervase Markham wrote:
João Miguel Neves
law is all rights reserved. So anyone that attempts to reuse must be
pretty sure (s)he has a license for the translation that allows her/him
to do it. Being in the public doesn't mean you can do whatever you want
with it.
Best regards,
João Miguel Neves
Simos Xenitellis
> No. You can't reuse the translations without an license. The default by
> law is all rights reserved. So anyone that attempts to reuse must be
> pretty sure (s)he has a license for the translation that allows her/him
> to do it. Being in the public doesn't mean you can do whatever you want
> with it.
Point taken.
Therefore, if there are any objections with the following paragraph,
please say so.
"This service includes the translation files of several free and
open-source projects. You may use the translation results of
a project (such as GNOME and KDE) as long as your project is licensed
with the same license. For example, both GNOME and KDE carry the GPL
license; therefore if your program is also licensed with the GPL license
you have the freedom to use this service. For a list of the licenses of
each project, see http://open-tran.eu/db.html [[Jacek may add an extra
column with distribution license, or it's up to the user]]. If you have
further questions, please consult the organisation behind your
distribution license."
Any objections?
Best regards,
Simos
Axel Hecht
> On Fri, 2007-03-30 at 17:16 +0100, João Miguel Neves wrote:
>> No. You can't reuse the translations without an license. The default by
>> law is all rights reserved. So anyone that attempts to reuse must be
>> pretty sure (s)he has a license for the translation that allows her/him
>> to do it. Being in the public doesn't mean you can do whatever you want
>> with it.
>
> Point taken.
>
> Therefore, if there are any objections with the following paragraph,
> please say so.
>
> "This service includes the translation files of several free and
> open-source projects. You may use the translation results of
> a project (such as GNOME and KDE) as long as your project is licensed
> with the same license. For example, both GNOME and KDE carry the GPL
> license; therefore if your program is also licensed with the GPL license
> you have the freedom to use this service. For a list of the licenses of
> each project, see http://open-tran.eu/db.html [[Jacek may add an extra
> column with distribution license, or it's up to the user]]. If you have
> further questions, please consult the organisation behind your
> distribution license."
>
> Any objections?
>
As Gerv pointed out earlier, we're not able to give you legal council.
From what I read on kde.org, the kde libs are under LGPL, the kde apps
are under GPL. That doesn't make your story simpler.
Note, any copyright holder can decide to take action if she or he thinks
that your use of the copyrighted material infringes her or his
copyrighted work, which, in the case of Mozilla, are all contributors to
the files. Mozilla cannot and will not take responsibility for your use
of copyrighted material.
I have no idea about the copyright holders for KDE, nor for GNOME. Nor
did I check the licensing scheme of GNOME.
I'll leave it up to Gerv to decide on whether he wants to issue a
Mozilla Foundation opinion on your text or not.
Axel
Gervase Markham
> You can choose only abide by the GPL. But if you want to contribute to
> upstream (Mozilla Firefox) then you must triple license it.
While this is true, it's not particularly relevant to the question we
are actually discussing, which is about making sure whatever is on the
site is correctly labelled :-)
Gerv
Gervase Markham
> Somewhat related question: how do you currently ensure that words used in
> Mozilla translations are not coming from copyrighted code? If a translator
> looks up a hundred (or hundreds of) words by comparing Windows versions of
> English and his language and includes them into Mozilla, would that not be
> a copyright problem?
Potentially. But I think it is highly unlikely that this is happening,
because all our translators are native speakers and so would find this
approach far more cumbersome than actually using their brains. :-)
Gerv
Gervase Markham
> "This service includes the translation files of several free and
> open-source projects. You may use the translation results of
> a project (such as GNOME and KDE) as long as your project is licensed
> with the same license.
For this sort of thing to work, each translation would need to be
labelled with its license, as I indicated in another message in this thread.
The users should not have to map project names to licences in their
head, when there's a perfectly good computer to do it for them with 100%
accuracy.
You should be storing licensing information along with each string.
Gerv
Gervase Markham
> What I see we need to write down here is what Jacek should exactly do so
> that we are all happy, projects and translators.
Well, really Jacek should be making sure he has the correct permission
for whatever use he envisages for the work of other people. :-) It's his
responsibility, not ours - and even if "we" are happy, doesn't mean
other copyright holders will be. The best guide is the text of the
licenses themselves, not us.
> I am trying to think
> how to make it comprehensive, yet usable by end users.
> I would really dislike to put excessive "restrictions" on open-tran.eu
> that would kill of the project.
Of course, but there must be some restrictions, because they come with
the text you got. If you take all your translations from projects under
the BSD or MIT licences, you would not have this issue.
> When I browser the Mozilla translations using LXR, I do not see any
> notification that the translation files are licensed, or even
> triple-licensed.
Indeed not. Which is unfortunate.
> Is this unintentional? I suppose someone could grab the translations
> from LXR and consider they are in the public domain.
No. The default, in the absence of a license, is "can't copy". So
everyone who uses our stuff at all is implicitly agreeing that it's all
under the tri-license, even if we haven't got the boilerplate in place
100% yet.
Gerv
Ricardo Palomares Martinez
> When I browser the Mozilla translations using LXR, I do not see any
> notification that the translation files are licensed, or even
> triple-licensed. For example, see
> http://lxr.mozilla.org/l10n-mozilla1.8/source/el/browser/chrome/browser/preferences/content.dtd
While it is true that many files show no license (due to a number of
reasons, one of them being that a fair amount of original en-US files
have no license themselves), there are also some l10n files licensed,
for instance:
http://lxr.mozilla.org/l10n-mozilla1.8/source/es-ES/security/manager/chrome/pippki/deviceManager.dtd
Hopefully, this lack of license headers will change for better. I
think that MozillaTranslator was the only L10n tool in use in Mozilla
environment not preserving the license headers, and that changed about
six months ago.
Anyway, this whole issue of licenses on translations (for this
particular project and at this particular moment) is kind of nonsense,
and let me explain before bashing me. :-)
Let's suppose I'm a translator working in a project not included in
open-tran.eu, and I know that "File" can be translated in spanish as
"Archivo" or "Fichero". So I decide to take a look at open-tran.eu to
see how "File" has been translated across all included projects in
open-tran.eu to see what would be the most consistent translation. I
even could get an additional translation that I didn't consider before
in the results list.
Now, how could anybody sustain a case against me if I choose *any* of
the results listed at open-tran.eu to use it in my translation? Those
translations are either wrong or "derivate" works not only of the
original projects, but also (and mainly) of the spanish dictionary,
because if they don't exist in the spanish official dictionary, they
can't be used in a right way in the first place.
The doubt could only arise if long en-US strings exist both in the
projects looked up by open-tran.eu and the project in which "I" (the
translator working in a project not included in open-tran.eu) am
translating. And, then, the problem wouldn't lie with open-tran.eu or
the translator using them, but with the two original projects.
I don't think that open-tran.eu is really doing a derivate work, but I
can understand that the original project licenses are mentioned of
otherwise considered in the website.
But the license headers in locale files (even in en-US files) only
have sense as a way to make legal guys happy, and maybe when someone
takes the whole files for other projects that probably also reuse the,
let's call it so, programmatic source code of Mozilla, because only
in that context could arise some kind of intellectual property. Words
and regular sentences (others than marketing or branding ones) can't
be subject to copyright issues, nor translations do.
Discussing if people can use the results provided by open-tran.eu is
pretty much like discussing if people could use "if",
"switch(discriminator)" or "for(i=0; i<list.length; i++)" just because
those character sequences appear in copyrighted source code made
public (and in this message, BTW). :-)
Ricardo
--
If it's true that we are here to help others,
then what exactly are the OTHERS here for?
Axel Hecht
Up to the point where you put a complete file in there. At which point
you're creating a derivative work of the original file and the other
files that are incorporated.
And this has been mentioned before by Gerv. One term is likely OK, 1000
are probably not, the line is drawn by judges.
Axel
Erdal Ronahi
> No. You can't reuse the translations without an license. The default by
> law is all rights reserved. So anyone that attempts to reuse must be
> pretty sure (s)he has a license for the translation that allows her/him
> to do it. Being in the public doesn't mean you can do whatever you want
> with it.
I am following the discussion with interest. I understand a lot of the
points raised here, but I think some considerations are missing.
What we have her is a kind of dictionary which draws from several
sources. The whole point in "reusing" the translations found here is a
process of reviewing existing translations from different projects,
which may notably differ from each other and then decide which of them
to adopt or reject. So there is much more in the process than just a
simple "copy & paste", it is also a creative activity of the
translator that makes use of this "dictionary". It may be a simple
choice when there are well established translations (for which noone
can claim a copyright) or need a lot of own creativity if none of the
existing translations are satisfying. But the whole process is very
different from, say, ripping a CD.
I think one should consider this also before discussing about possible
"copyright infringements". Therefore I find the discussion a bit
overheated.
Regards,
Erdal
Ville Pohjanheimo
Unless we're talking here about totally new words/jargon by the
translator, then the terminology has to have been established by other
programs. Whether the translator wants to or not, (s)he must in a sense
infringe if the program is to be understood by normal people....
It should also be noted that the same applies to en-US strings, thought
I suppose Mozilla is fortunate to have inherited Netscape's inventory of
original jargon. My guess is that a lot of the strings in use have been
invented under licences that are not compatible in Mozilla's current
scheme, in en-US that is.
Of course the "1 vs 1000 strings" point made by Gerv (and others) holds,
so these infringements might not be /actual/ infringements.
Then again... if copying one string is likely OK, why should each string
on open-tran.eu be served with a licence if only a group of strings is
protected? .. Uh, sorry for rambling. I just find some of this quite absurd.
-ville/mozilla.fi
Jacek Śliwerski
discussion. It was very interesting to read all your points of view on
the issue.
Gervase Markham wrote:
>
> I promise you, it's not. I would not be having this conversation with
> you if I thought that what you are doing is legal.
You think my project is illegal and I am of a different opinion. This
discussion is not going to change anything about it. I don't want you
to promise me anything more. If you are brave enough to protect the
OS/FS community from people like me - please, file a complaint against
open-tran.eu.
I appreciate that you are defending the rights of the creators of the
open-source software, but in my opinion you are pushing it too far and I
can't see any benefits for the open source community from it. I haven't
used the translations of open source projects in an non-open, non-free
software and I am giving you the option to remove the material, you are
calling infringing, from my database.
> You need to make your users aware of the license attached to each string
> you provide.
>
> For this sort of thing to work, each translation would need to be
> labelled with its license, as I indicated in another message in this thread.
This is going insane. I am not going to pollute the results of the
search with an information about the license terms of each of the
phrases, just as Google does not show the license terms of each website
on their results pages.
However, I have included the (slightly reworded) notice proposed by
Simos, which you can find at http://open-tran.eu/db.html.
Best regards,
--
Jacek
Robert Kaiser
> Words
> and regular sentences (others than marketing or branding ones) can't
> be subject to copyright issues, nor translations do.
Erm, books are. University papers are. Even newspaper articles are. So
they _can_ be subject to copyright issues. Copyright usually dictates to
at least mention the source of something you use or cite from some
other's writings. Even on excerpts (sentences or parts of sentences).
Robert Kaiser
Robert Kaiser
> You think my project is illegal and I am of a different opinion.
I think you misread him. Your project is a very, very good idea and NOT
illegal by itself. It's just very easy to use it illegally, and you
should help prevent that.
>> For this sort of thing to work, each translation would need to be
>> labelled with its license, as I indicated in another message in this
>> thread.
>
> This is going insane. I am not going to pollute the results of the
> search with an information about the license terms of each of the
> phrases, just as Google does not show the license terms of each website
> on their results pages.
This is not insane, actually, it's just reality. Google satisfies
Copyright with directly linking and pointing to the source of the text
it cites. If you're annotating which project the source was found in,
that should be legally enough to tell any lawyer that the responsibility
of following that project's licensing lies in the hands of the
localizers reusing that string.
If you want to help that localizer even more, you have some easy-to-find
source (e.g. linked from every page, probably from a line telling "Be
sure to follow the licenses" or such) for people to find out which
project is using which license for their localizations.
Again, the idea is really great, having such a database surely helps
lots of people, and I'm looking forwarded to seeing SeaMonkey included
once it's got its localizations in Mozilla's CVS.
We'd only like to prevent legal problem for your users or probably
yourself before they can arise.
Robert Kaiser
Robert Kaiser
> Unless we're talking here about totally new words/jargon by the
> translator, then the terminology has to have been established by other
> programs. Whether the translator wants to or not, (s)he must in a sense
> infringe if the program is to be understood by normal people....
>
> It should also be noted that the same applies to en-US strings
True, the whole topic is rather complicated and fragile legally. Knowing
that there have been fights over using the common word "windows" as a
trademarked name, and people fighting their copyrighted writings being
reused by others (without properly identifying the source), this gets
even more complicated and fragile.
I don't think we can even grasp what is really legal or not in this area
as long as no court has have made a decision over a case on such a
potential infringement.
On the other hand, this actually might be the reason why IE has
"Favorites" instead of "Bookmarks" and similar stuff.
Robert Kaiser
F Wolff
I would find it unlikely if this is _not_ happening. Don't all
translators use TM functionalities? Both msgmerge and pot2po allows this
(sorry - I'm not familiar with MT to comment on that).
I will just insert some of my thoughts here. They don't necessarily all
belong here in the thread, but it has grown a bit big now anyway :-)
A translator would own the copyright on all translations they have done
for Free software projects, but many people would consider those to be
derivative works of the original English strings in the software. In
such a line of reasoning, some people might say that translators are not
allowed to use a TM of Evolution (GPL) translations to translate
Thunderbird (moz tri-licence suite). In the case of open-tran.eu, the
redistribution right is not because of copyright ownership, but only
because of the licence anyway.
Considering what the purpose of the translation is, of course it is good
to reuse these translations as much as possible. Consistency makes
software easier to use. So we translate in the same way for the same
reason that the English strings correspond to other pieces of software.
Now if someone makes a new dictionary by researching several existing
(normal all rights reserved copyright, not copyleft) dictionaries, they
are allowed to publish a new dictionary, even if it contains many
resemblances to all the reference dictionaries. Somehow this has never
seemed to be a problem, even to claim full copyright of the newly
created dictionary.
I guess we all agree on the case of a dictionary lookup - a word can't
be copyrighted; the relationship with the English (say) of a bilingual
dictionary can't be copyrighted either. The copyright is on the
collection - the creative work. Of course, we also get expression
dictionaries (dictionaries with translations of multiword expressions,
like idioms), and by extension these individual entries then can't be
copyrighted either - they are considered basic building blocks of the
language. Now both these cases are already presented in a TM of Free
software translations: word-word relationships as well as small
multiword expressions ("Are you sure you want to ...."). It is
reasonable to expect frequent users of translated software to simply
know what the most idiomatic translation is (perhaps only true for
languages with a long history of localisation like de, fr, es, etc.)
Now, consider fuzzy matching. If open-trans.eu gives me a fuzzy match of
a translation (say 70% similarity) and I have to rework it so that it is
correct, what is the legal situation here? open-trans.eu had the right
to distribute the text to me - it was under some copyleft licence (let's
ignore the fact that I'm supposed to know the exact terms). I had to
adapt it. We could call it a derivative, but all it did was to save me
time and to ensure some measure of consistency with something. I could
have redone it and in many cases have reached a similar result. Is the
final translation a derivative of the English, or of the TM match? Both,
but is this strong enough that the "viral" clause of copyleft licences
kick in? I could have looked up the expression in an all rights reserved
expression dictionary, and no-one would say I'm not allowed to use it,
not so? I could do it for each expression in my untranslated file that I
can find in the copyrighted dictionary. Why then different for
open-trans.eu?
I guess my case is that what open-trans.eu does can probably be seen as
a _use_ of the licenced material, rather than redistribution. Of course,
it is distributing minute parts of it, but it is done so as part of a
new work, as much as the original. As soon as we consider this to be use
and not redistribution, the whole picture changes.
Now, things like trademarks and client confidentiality (if we add TM use
of proprietary translations to the soup) is something entirely
different, and I believe that these should probably be considered
separately, but in practise these are probably not real issues.
(epen-trans.eu is not using confidential stuff, and trademarks are
unlikely to survive in a cycle of fuzzy matching -> fixing -> review.)
Friedel
Gervase Markham
> What we have her is a kind of dictionary which draws from several
> sources. The whole point in "reusing" the translations found here is a
> process of reviewing existing translations from different projects,
> which may notably differ from each other and then decide which of them
> to adopt or reject. So there is much more in the process than just a
> simple "copy & paste", it is also a creative activity of the
> translator that makes use of this "dictionary".
That is quite possibly true. It's also irrelevant to the question of
whether any particular use is copyright infringement.
If I get five novels by Philip K. Dick, and combine selected paragraphs
from them to make a new novel with a different plot, the result is a
derivative work of all five, not a derivative work of none of them, even
though I also had creative input in choosing which paragraphs and in
what order.
Gerv
Gervase Markham
> I think you misread him. Your project is a very, very good idea and NOT
> illegal by itself. It's just very easy to use it illegally, and you
> should help prevent that.
Indeed. This is what I meant; I apologise if I wasn't clear.
>> This is going insane. I am not going to pollute the results of the
>> search with an information about the license terms of each of the
>> phrases, just as Google does not show the license terms of each
>> website on their results pages.
You "pollute" the results with a long list of filenames already. Ditch
the filenames; insert the license terms.
> This is not insane, actually, it's just reality. Google satisfies
> Copyright with directly linking and pointing to the source of the text
> it cites. If you're annotating which project the source was found in,
> that should be legally enough to tell any lawyer that the responsibility
> of following that project's licensing lies in the hands of the
> localizers reusing that string.
> If you want to help that localizer even more, you have some easy-to-find
> source (e.g. linked from every page, probably from a line telling "Be
> sure to follow the licenses" or such) for people to find out which
> project is using which license for their localizations.
As Kairo points out, the results of a Google search are not presented
specifically for the purpose of copying and pasting them into something
else. And if someone were to do that, it would be copyright infringement.
Again, I repeat: showing someone a list of ways a word is translated is
not necessarily copyright infringement - but as soon as someone uses
those phrases in a way that you are encouraging them to use them, it is.
So you should make it clear, as a service to your users, what terms they
can legally use the translations under.
Again, I repeat: open-tran.eu is a great idea, but you should make it
clear to users what the legal situation is so they don't break the law
by mistake. They won't be happy if they do, and get in trouble.
Gerv
Gervase Markham
> Even for people with brains it often is difficult to decide which
> particular word to pick if there are several possible ones to choose
> from. In my past translation work I then looked up what the system and
> perhaps other progrems were using, for consistency or to double ckeck.
> That may not have happened for hundreds of words but for several tens
> (not for Mozilla work, so you are safe). Until reading the copyright
> part of this thread I thought this would be legitimate use...
There is a difference between inspiration and copying; for one-or-two
word snippets, the practical difference comes down to whether you used a
copy and paste tool, or retyped the word.
Like I said earlier, no-one can "copyright" the idea of translating File
as Fichier, but they can copyright their instance of that translation so
that you may only _copy_ it under their terms.
Gerv
Gervase Markham
> I would find it unlikely if this is _not_ happening. Don't all
> translators use TM functionalities?
"TM functionalities"? What does TM stand for? I only know it as
"trademark". And our discussion has nothing whatsoever to do with
trademarks.
> A translator would own the copyright on all translations they have done
> for Free software projects, but many people would consider those to be
> derivative works of the original English strings in the software.
They are derivative works, so the copyright is joint.
> In
> such a line of reasoning, some people might say that translators are not
> allowed to use a TM of Evolution (GPL) translations to translate
> Thunderbird (moz tri-licence suite).
One should not be copying and pasting from Evolution translation files
into Thunderbird ones, no. (Unless you are the person who originally
wrote the Evolution files, or have gained permission from the people who
did.)
> Considering what the purpose of the translation is, of course it is good
> to reuse these translations as much as possible. Consistency makes
> software easier to use. So we translate in the same way for the same
> reason that the English strings correspond to other pieces of software.
Of course.
> Now if someone makes a new dictionary by researching several existing
> (normal all rights reserved copyright, not copyleft) dictionaries, they
> are allowed to publish a new dictionary, even if it contains many
> resemblances to all the reference dictionaries. Somehow this has never
> seemed to be a problem, even to claim full copyright of the newly
> created dictionary.
Do you mean "dictionary" as in "list of words" (like a spelling checker
dictionary) or "dictionary" as in "list of words with their definitions"
(as in a book dictionary)?
For the latter, the obvious copyright and creative step is in the
explanations of what the words mean.
> Now, consider fuzzy matching. If open-trans.eu gives me a fuzzy match of
> a translation (say 70% similarity) and I have to rework it so that it is
> correct, what is the legal situation here? open-trans.eu had the right
> to distribute the text to me - it was under some copyleft licence (let's
> ignore the fact that I'm supposed to know the exact terms). I had to
> adapt it. We could call it a derivative, but all it did was to save me
> time and to ensure some measure of consistency with something.
That doesn't make it not a derivative. Lots of people make derivative
works precisely to save time and achieve consistency.
> I could
> have redone it and in many cases have reached a similar result. Is the
> final translation a derivative of the English, or of the TM match? Both,
> but is this strong enough that the "viral" clause of copyleft licences
> kick in?
Good question. As I've said earlier in the thread, ask a judge. No-one
is going to take someone to court over one string - but my concern is to
help people not to break the law, not to find ways to sue free software
developers!
> I could have looked up the expression in an all rights reserved
> expression dictionary, and no-one would say I'm not allowed to use it,
> not so?
I think that if a dictionary author sells you a copy, there's an
implicit right to look stuff up and use the result.
> I guess my case is that what open-trans.eu does can probably be seen as
> a _use_ of the licenced material, rather than redistribution. Of course,
> it is distributing minute parts of it, but it is done so as part of a
> new work, as much as the original. As soon as we consider this to be use
> and not redistribution, the whole picture changes.
It's both use and redistribution. How else does it get to my PC so I can
view it? And as soon as someone copies and pastes a translation into
their file, it's certainly redistribution.
Gerv
Ricardo Palomares Martinez
> Ricardo Palomares Martinez schrieb:
>> Words
>> and regular sentences (others than marketing or branding ones) can't
>> be subject to copyright issues, nor translations do.
>
> Erm, books are. University papers are. Even newspaper articles are. So
> they _can_ be subject to copyright issues.
But 'they' as in the book, the university paper or the newspaper
article as a whole, or a significant part of it. Claiming copyright
issues in a regular sentence that can be easily found in previous
artwork not subject to copyright, just because it is included also in
the protected artwork has no point.
http://www.chillingeffects.org/piracy/notice.cgi?NoticeID=4096#QID459
> Copyright usually dictates to
> at least mention the source of something you use or cite from some
> other's writings. Even on excerpts (sentences or parts of sentences).
I've been reading some legal articles on copyright infringements and
"fair use" doesn't seem to oblige to such mention (which I also
honestly expected to be needed).
Anyway, I really think this discussion has no point because:
1) How can Mozilla, KDE or GNOME claim that a copyright infringement
has happened (by means of open-tran.eu or not) based on translations
without previously claim a copyright infringement on the original
en-US strings?
2) Except in long sentences (which undoubtedly would let us back to
#1), how can anyone prove that /my/ "File" -> "Archivo" translation
covered by MPL/LGPL/GPL has been used by Microsoft or the reverse? And
the point here is not so much the "prior art" as the translation accuracy.
Ricardo.
Axel Hecht
You're missing a whole bunch of points here mentioned elsewhere in this
thread.
> 2) Except in long sentences (which undoubtedly would let us back to
> #1), how can anyone prove that /my/ "File" -> "Archivo" translation
> covered by MPL/LGPL/GPL has been used by Microsoft or the reverse? And
> the point here is not so much the "prior art" as the translation accuracy.
Again, open-trans offers translations of complete files, at which point
it creates copyrighted derivative work, and probably even grabs
differently licensed content together. Mentioned before, too.
Axel
Samuel Murray
> F Wolff wrote:
> > I would find it unlikely if this is _not_ happening. Don't all
> > translators use TM functionalities?
> "TM functionalities"? What does TM stand for?
Translation memory. I find it strange that some localisers are not
familiar with the concept of TM. TM is used in all kinds of
translation, but the type that is most likely to benefit from it,
surely, is localisation.
> > I could have looked up the expression in an all rights reserved
> > expression dictionary, and no-one would say I'm not allowed to use it,
> > not so?
> I think that if a dictionary author sells you a copy, there's an
> implicit right to look stuff up and use the result.
True, but you'd still be reusing a piece of proprietary "code" in a
software product that disallows proprietary pieces of code in it. The
fact that the dictionary author had licensed you to use his code
doesn't mean he had licensed you to relisence it. If I were to follow
your reasoning, it would mean that translators may use a dictionary
only if they do no intend to redistribute their translation.
I think the issue here is building blocks. At some point in reducing
code to its basic building blocks, you'll find a block so small that
it is generic, and not subject to copyright.
I can take an HTML page from the web and copy the following piece of
code from it:
</head><body>
and my "copy" of it will surely not be regarded as a derivative work
of that HTML page. This is because what I've copied is generic.
The same can be said of pieces of text. They are building blocks of a
larger whole, and although the whole is subject to copyright, at some
point the smaller units must be excempt from copyright because they
are not specific enough to be copyrighted.
If I take all the Mozilla strings, and I reduce it to a long list of
unique words sorted alphabetically, is my list a derived work of
Mozilla? I say, no. What if I were to reduce it to a long list of
unique characters, sorted alphabetically? I'd end up with a list that
is less than 100 entries long... is my list a derived work of
Mozilla? You'd be silly to believe it, I think. The question is... at
which point do the building blocks become so large that they are no
longer generic. Are words generic? Yes. Are phrases generic? I
think so. Are whole sentences generic? I would hope so.
Let's get back to my HTML page example. If the author of the new HTML
page that contains that piece of code copied from the web, freely
admits that he had taken that piece of code from that web site... can
he be charged with copyright infringement? I think that even if he
freely admitted his act, it would not be regarded by any court as
copyright infringement.
Your comment?
Samuel (aka leuce, aka voetleuce)
Ricardo Palomares Martinez
> Samuel Murray wrote:
>> If I take all the Mozilla strings, and I reduce it to a long list of
>> unique words sorted alphabetically, is my list a derived work of
>> Mozilla? I say, no.
>
> created your own identical-looking list, it would not be.
And could Mozilla Foundation/Corporation prove that such result list
has been done in either way except because the offendant voluntarily
admits it?
I understand that no Mozilla representative can really bless the usage
of open-tran.eu if legal implications of current licenses dictate the
reverse, even though the implications are as absurd as telling me that
I can't copy & paste from open-tran.eu window to my L10n tool/UTF-8
editor window but I can tile both windows in my screen and retype what
I see in one of them to other; remember that I'm seeing a bunch of
results in open-tran.eu, so I'm not specifically retyping any specific
instance of them.
But if open-tran.eu can't be used (wow, how dangerous is to mistype
"used" here) :-) in that way, I agree with Samuel and Friedel in that
translation memories, which are in wide use in PO-based tools, would
fall in the same kind of copyright infringement. And I tend to believe
that most locale teams that have worked in Firefox 2 use Translate
Toolkit and PO-tools (MozillaTranslator doesn't include translation
memory from external sources yet), so we could end renaming this
thread to (paraphrasing another hot one in this ng) "From 50 locales
to 10". And I'm not missing a "0".
Really, if Mozilla is so serious about this issue and want localizers
to strictly adhere to this whole licenses and copyright bounds, it
will be really difficult to attract localizers (often working in other
projects besides Mozilla), which felt their time is wasted if they
have to re-translate the exact same words and sentences just because
KDE uses GPL and Mozilla MPL. Mozilla should think if there is any
kind of solution that allows and encourages re-use of translations
across differently licensed projects that have in common their
open-source nature.
Besides that, I'm curious to know if the original poster and the one
in charge of open-tran.eu, Jacek Sliwerski, has done similar
announcements in KDE or Gnome mailing lists/newsgroups, and if similar
legal issues have been raised there. After all, similar concerns could
be made by KDE/Gnome projects regarding the ability of a closed-source
project to search open-tran.eu and copy&paste the results in the
closed translations.
Ricardo
Jacek Śliwerski
>
> Besides that, I'm curious to know if the original poster and the one
> in charge of open-tran.eu, Jacek Sliwerski, has done similar
> announcements in KDE or Gnome mailing lists/newsgroups, and if similar
> legal issues have been raised there.
I have posted similar announcements on kde-i18n-doc and gnome-i18n.
Neither GNOME (I have been corrected to spell it this way, because it is
an acronym, actually), nor KDE translators have raised any legal issues
regarding the copyright. Moreover, I have received many requests from
private people and even one company, to include their localizations in
my database.
Best regards,
--
Jacek
Gervase Markham
> Neither GNOME (I have been corrected to spell it this way, because it is
> an acronym, actually), nor KDE translators have raised any legal issues
> regarding the copyright.
That's quite possibly because many people don't bother to think about
this sort of thing. The Mozilla project spends a lot more time thinking
about licenses than the "everything is GPLed" world.
The fact that they have not raised issues doesn't mean there aren't any.
> Moreover, I have received many requests from
> private people and even one company, to include their localizations in
> my database.
Well, if they own the copyright, they can do what they like with it,
including giving it to you to put in your database and serve to the
world :-)
Again, for the Nth time, I think it's a great idea, but I think you
should do it in a way which does not encourage others to commit
copyright infringement.
Gerv
Jacek Śliwerski
>
> The fact that they have not raised issues doesn't mean there aren't
> any.
Nobody has implied that. I just wanted to satisfy Ricardo's curiosity.
My email contains only dry facts, no judgments.
In contrast to you, I leave it to the readers, to make up their minds
about what is right and what is wrong.
> The Mozilla project spends a lot more time thinking about licenses
> than the "everything is GPLed" world.
And it seems to make you feel better ;]
Well... while you are thinking about licenses, the "everything is GPLed"
world is developing software.
Best regards,
--
Jacek
Robert Kaiser
> Well... while you are thinking about licenses, the "everything is GPLed"
> world is developing software.
Now you are insulting the Mozilla project.
I don't think that takes any of us far.
We could also say "while we are thinking forward, you're leading people
into breaking laws" - but I at least won't say that because it would be
as insulting to you as your statement above is to us.
Developing software and thinking about its legal implications are
orthogonal (and, for the most time, probably even done by different
people in our project).
Note that not thinking about its implications even got some GPL-using
people sued for breaking laws.
Robert Kaiser
Simos Xenitellis
I think the phrase missing here, and most of the thread, is "be
constructive". I have been following the thread and it is so slow to
reach a result/conclusion.
I sent a few emails on "okay folks, what should *actually* be done", and
I got lukewarm response to suggestions. It is so easy to criticise and
shoot down projects.
I do not see much development to this thread; if anyone really cares
(Mozilla), please contact the Foundation lawyers on the legal standing
of translations, and how these translations can interact with
translations from other projects.
Simos
Robert Kaiser
> I think the phrase missing here, and most of the thread, is "be
> constructive". I have been following the thread and it is so slow to
> reach a result/conclusion.
It's been clearly said that the only possible solution is annotating the
license of every string with the string, either directly or via noting
which project is comes from and have a clearly visible link that tells
which license(s) the strings of every project are under.
And, BTW, that is what even the licenses themselves require (!) one to
do there.
If any string I have directly contributed is put up there without the
license of it being mentioned, I'm inclined to put any reasonable effort
into either getting the string removed or the license correctly noted.
Note that there are probably already strings up there that I have
indirectly contributed (i.e. that my colleagues in the German team
copied from my suite localization, not to speak of the few English
strings I may have helped with).
Robert Kaiser
Gervase Markham
> I sent a few emails on "okay folks, what should *actually* be done", and
> I got lukewarm response to suggestions. It is so easy to criticise and
> shoot down projects.
I have suggested what should be done multiple times in this thread. I
have gone out of my way to be constructive. As I have said several
times, I think this is a good idea.
> I do not see much development to this thread; if anyone really cares
> (Mozilla), please contact the Foundation lawyers on the legal standing
> of translations, and how these translations can interact with
> translations from other projects.
There's nothing special about translations - they are just like any
other code. And the interactions between code under different licenses
is a well-understood issue. You can't, for example, take GPLed or MPLed
code and paste it into a BSD-licensed project. Yet open-tran.eu does not
warn people who might be about to do just that.
Gerv