Rick, James, and Aleksey, thanks for your comments.
James Pritchard jnpritchard@btopenworld.com (2004/02/17 23:56:35) I like this idea, it will encourage testers/users to get involved with reporting problems. But as is said, we need a working TCP/IP networking system before it will work. I think pressing the button could do 1 of 2 things. Either start a web browser window that loads bugzilla or our own problem tracker; or it launches a little app that asks a few questions. It would be passed the calling apps name and version etc when it is executed. The all the app needs is space for the reporters name, email and a box for a description of the problem. The most important thing is to keep it simple.
Thanks. I think the latter's the better, I'll explain the why later.
As for how the translations are done. There should definatly be a stage of review/evaluation in my opinion. So the translators submit translations to a coordinator for each language that checks them before passing them onto Aleksey for committing to CVS.
Yes, it should be; as long as the language have at least two translators. I'll also mention about this above as a reply to Aleksey.
"Aleksey Bragin" aleksey@studiocerebral.com (2004/02/18 00:56:22)
I read the discussion on translation on this list and thought that almost the same problem as other OSS have'll happen on ROS translation, too. IMO, you can handle those languages which have many translators easily(in this case, German, Dutch, and some other western languages).
I may sound obvious but my point is to not let this happen.
Well, every translation coodinator wills to do so, but unfortunately, it's quite difficult matter to deal with. I'll also explain about this on the later paragraph. (Don't take me wrong! I'm not saying that you're not capable to take the role or something.)
If there're strange translations, they are easy to be found and fixed. When there's a argument about translation, it's also solved easily; you can vote or do anything you like. But on the languages which have few translators, the circumstances'll be definitively different. Before
Yes, I agree with you --- if a language have only one translator -- than I'd better postpone the translation until we have at least two, who can review each other's work. Otherwise the situation you're describing will certainly happen. (Replace 'at least two' with 'at least n number of people' - where n
= 2).
So, will the users who can only understand their own languages make to wait for the appearance of another translator despite there're already translations ready? Well, a bit evil question;) It's not the real point that I want to discuss here.
<snip; refer the former mail>
I don't know how other people likes this, but for now unfortunately I think it's not very suitablel. Despite ROS is just technically can't do this (e.g. lack of TCP/IP) - but even if we can, imagine very common situation - ROS bugchecks arfter user have tried to run some application. And what to do? No dialogs, nothing except for the blue text-screen showing stack trace. Or another error example --- Quake-II is shown as bottom-up. User can go to Bugzilla on the website and report this error. Or can click on that 'report bug' button (unfamiliar to Windows user at all, and testers should know bugzilla already, with these buttons) and still write the same text, and send it using ROS (when it becomes mature enough). In this case bug-report button is more suitable, but I still think it doesn't worth implementing it yet.
Well, unfortunately, I think you and James have missed the point. I'm not suggesting to implement the feature to mainly enhance "testers'" comfort on reporting bugs, but to make the general users into the testers. I asked the admin about the number of ID's registered on reactos.com, and the answer was 551 atm. 550, twenty times smaller than the number of reactos-0.2.0 ISO image downloads. (Strictly, comparing these numbers might be nonsense, because about 10% of the registered users have "used" the account to report-bug/post-on-forum or something, on the other hand, not all people who have downloaded the ISO should've tried on it.) Then, let's use the number for explanation; 1/20 of the users've registered on reactos.com, 19/20 users didn't. General users don't like registering on the site only to report a bug or comment on translation. They want themselves to be anonymouse, we should grant the fact. Some of you may suspect the anonymity mess things up to dumb. But AFAIK, it's not true. Some Japanese forums are running under almost full anonymity, most of the users post with default name (Imagine the situation; most of the users on slashdot post their comments as Anonymouse Coward.) Of course there're noise, but most of the users wisely ignore them. I think we can adapt this model on bug-reports(tho I must admit that there's a cultual diffence between eastern countries(now especially about Japanese) and the western countries). The number of anonymouse users who complained about some features/translation, it'll help developers/translators to make priority-based well-constructed TODO list, because it's something like a "vote and/or comments about enhancement". At least for the translation, it's a speed/quality tradeoff, you know. I guess you(Aleksey)'ll consider much about quality rather than quantity. So, the well-constructed TODO list is the must, because the translation'll take much time and consideration; I mean, you should translate following the context and you can't do it well without previewing the messages in the applications.
Then, please take a look at KDE's i18n process(sorry if you've already read). They have about five years of knowhow which are well-documented as kBabel, FAQ, and tutorial. I saw many problem they dealed with on the process of i18n. They'll be a great help on ROS i18n/l10n. http://i18n.kde.org/ *Note* even now, even they have average of tens of translators/language, 1/3 of the languages are at enough quality on GUI stuff. See the chart below: http://i18n.kde.org/stats/gui/HEAD/index.php On the other hand, about documents: http://i18n.kde.org/stats/doc/HEAD/ .......well, let's forget about the documents. They can't be simply translated as same as the GUI stuff(there are numbers of unofficial documents).
But I'd like to repeat; remember, many users complain the terrible quality of translation but curerntly, most of them just complain, overlook, and spread bad image of those softwares, but don't report and try to make them better. It'll be a great help if we can take them into the project as testers.
hmm.. I think if the translation is really bad, then really a lot of people will notice that. And if someone is spreading the bad image of the software, it might be the one who doesn't like that software at all, translation is just an excuse :-) (joke)
hehe, true.
Regards, Masahiro.