Re: [gnome-cy] Kartouche, CVS, keyboards, etc.

On Mon, Mar 17, 2003 at 11:29:27AM +0000, Kevin Donnelly wrote:
> > - Links are the same colour as other text. This is a big web design no-no.
> Yes.  Unfortunately the default link colour doesn't go with the
> schema, and underlines even less so.  I'll have another look at the
> stylesheet and see if I can do something to address this.

That should be fairly straightforward.

> > - It would be nice to have column headings ("Original", "Current
> >   Translation", "Current Suggestion")
> OK, I'll add that.


> > - It would be nice if the checkboxes could go away, which could be done
> >   by ignoring all results without non-whitespace (!/\S/) contents.
> >   Implementing this, and the previous suggestion, would make the
> >   interface immediately less confusing, I think, especially for people
> >   like me who start playing around without reading the documentation
> >   first.
> Hmm.  Not so sure about this one.  I have this paranoia that when you
> make something automatic to save yourself work you go back a couple of
> days later and find it has eaten your cat.  But I can see that it
> would streamline things.  I'll do a few tests on this.  Would you like
> to contribute a regex you think will be foolproof (and I'm the fool in
> question, remember)?

Well, assuming PHP's regexes are similar enough to Perl's the string
should be used if it matches /\S/ (i.e. it contains at least one
non-whitespace character).

> > - What is the Specific Item thing? It complained when I tried playing
> >   with it and I didn't see it documented anywhere.
> Oh, I like this, it's always very well-behaved with me.  Suppose
> you've translated up to string 39 in a file.  The next day you go back
> and need to start from string 40.  You just type that into the box,
> press the button, and viola!  You can achieve much the same effect
> with the "show only untranslated" switch on the page where you select
> the tables, which I added last week, but that is more for winkling out
> those last few untranslated strings that are lurking in corners.

Ah, perhaps the Specific Item thing should be described on the Usage
page then. Now that I understand it, I might start using it.

> > - It would be nice to have a list of pages, possibly in the form of a row
> >   of links in the form:
> >   Jump to Messages: 1-20, 21-40, 41-60, 60-67
> Yes, Chris Jackson also asked about this, but the Specific Item and Show 
> Untranslated address this need.  The main reason why I didn't implement this 
> is that some of the files are quite big, and if you had a list of page 
> numbers you would have a ridiculous amount of them in these cases (eg
> 150 for kdelibs).  

Good point.

> > - It would be nice if there was a working "show untranslated only" mode,
> >   with correct paging, so that it's very easy to start filling in gaps
> >   without having to wade through stuff that's already done. The page
> >   list would have to cooperate with this. (That is, each page should
> >   contain 20 untranslated messages, not just the untranslated messages
> >   with indexes within N to N + 20, if you see what I mean.)
> Yes, this is implemented if you use the "show untranslated" switch (see 
> above).  Beat you to it on this one, eh?


> > - More statistics would be nice, both on the Start Translating and the
> >   Hall of Fame pages: specifically, number of strings translated,
> >   and number of suggestions pending approval.
> Well, funnily enough, on the admin bit I have implemented a
> "suggestions pending" column, and I was just about to transfer it to
> the userland page.  So that's OK.  What did you have in mind for the
> HoF?  I was also wondering about implementing a "suggestions submitted
> today" page, which will hopefully 

For the HoF page, I was envisioning something along the lines of

Who	Suggestions	Suggestions Pending	Accepted Translations
Bleddyn	103		32			65
Owain	43		0			39

> show an exponential increase.  I'm going to look at that shortly.

Well, a graph would be very nice :). And actually might be very useful,
if it shows Number of Strings and Number of Strings Translated over
time, since, of course, the Number of Strings /will/ change and
translation is ongoing in nature rather than a one-off event.

> > I might have been more inclined to actually implement some of these myself,
> > but I'm afraid I harbour a strong dislike for PHP.
> What?  Do my ears deceive me?  For why?  PHP is *so* cool, because
> twits like me can actually get something running very quickly, and
> it's very forgiving.  It's also actually pretty reliable.  Do you do
> Perl, or what?  If so, beware, because I may send you queries about
> regular expressions ......

I've just had a few mishaps with it in the past. Yes, I do use Perl, and
yes, I'm happy to answer any questions you might have about regular
expressions as best as I can.

> > I found the Omnivore very helpful, and it would be even better if it was
> > fed translations made via Kartouche, since it would help consistency as
> > well as expanding the number of things it knows about. 
> Yes, it is actually amazingly useful - I didn't think it would be.  I
> have now worked out a simple way to add the KG/100 translations (eg
> the translations in all the 100% files were swept in last night, so we
> now have nearly 8,000 entries there now).  At the minute this is done
> manually, because I need to add another field to the tables, and I
> don't want to mess things until I've tested it, but since all it
> requires is ticking a box and pressing a button, this is not
> difficult.  I want to add Rhoslyn's MDK files today, and Dewi's Moz
> files later (they will be based on the downloadable web-page, so they 
> will not be perfect - in particular, I've noticed some of the longer entries 
> are truncated).  I will also add the file available at,
> but this has only about 800 general words, and some of them are a bit
> suspect.  Still, all grist to the mill.

Yes, I noticed the import some time this morning :). This makes it even
more useful since you can instantly get an answer for "how have other people
translated this word already?"

> > Maybe it should even keep track of failed searches since they will
> > probably indicate where there are gaps. 
> Yes, I had dismissed that, but I've just thought of a simple way it could be 
> done.  The slight problem is that to get the best out of this, you'd need to 
> tie the Kartouche front-end onto Omnivore, which should be easy, but would 
> take a bit of time to sort out.  So the ability for users to add to Omnivore 
> itself probably won't appear for a while.  But then the key aim is a Welsh 
> desktop, not a Welsh dictionary, so that's probably OK.

I don't understand the need to link Omnivore with Kartouche here. I was
just envisioning a simple mechanism where searches would be logged to a
file if they didn't return any result, and somebody could look in the
log at lesiure to see what they can see. On the other hand, this may be
a non-feature.

> > Oh, and also: can it not open in a new window? To my mind, if people
> > want a page in a new window they will take the trouble of asking the
> > browser for it explicitly, and people who want it to open in a new
> > tab have to ask for that explicitly anyway.
> Ah, interesting.  The way I use it is to have a Kartouche window, and
> an Omnivore window, because it really bugs me when I click on a link
> and it eats my existing window and I have to click Back to get back to
> it, and then the Omnivore window has disappeared again, and then I
> have to get *it* back, and so on.  I know new windows can confuse
> people ('ere, where's me translations gone? I can't click back ...) so
> that's why I flag the new window.  It would be even better if you
> could get IE and Moz to open at a smaller size (Konqueror does this,
> but I can't see where you would set the other to to do so - in the
> HTML possibly - I shall have to look at that) - that would make 
> it clear to people that this was just an adjunct to the main KG/100 page.

Right, what I do is that I start using Kartouche, and then as soon as I
need an Omnivore window I open it in a new tab and switch back and forth
as need dictates. This is very efficient as I can keyboard-naviagate
back and forth very quickly.

> > Again, I have to say that, on the whole, I am very pleased that Kartouche
> > exists and like it a lot. I'm also very happy that the KDE and Gnome
> > efforts are coordinating with each other, since this can only make our
> > respective projects easier and our results more consistent.
> Good :-)  It's certainly doing wonders for my tech Welsh - I'm going around 
> now saying "rhaglennig" to everyone.

Excellent :).

> > I'm a little fuzzy on the relationship between Kartouche and Gnome
> > CVS. Are the strings in Kartouche taken from HEAD? How are new
> > versions of messages handled? How do strings get from Kartouche into
> > CVS? It would be nice if CVS were not far behind Kartouche, since it
> > means that the GTP's statistics will be up-to-date.
> Well, I received my Gnomeries from the estimable Mr Cox, so I can't
> say, I'm afraid.  I intend in the next couple of days to send him the
> 100% Gnome files to do whatever is best with them (although I may keep
> them on KG/100 for a while pour encourager les autres).  I am hoping
> he will not come back and say "I've just discovered a show-stopping
> error in the exports you sent me" .....

I definitely agree that "finished" files should be kept around for a
while... in fact it may not be worth taking them out. At some point, our
copies of the English string lists are going to become obsolete, at
which point we will want to merge the translations we've got with the
translations we haven't somehow. I suppose it's something along the
lines of taking the new English file, inserting existing Welsh
translations made against the old English file my matching msgids, then
dumping the result in the database.

> I am also trying to read up on CVS myself, so that I can start a tree for 
> Kartouche - I have a testing version, a deployed version, and a downloadable 
> version, and it's beginning to get a bit difficult knowing what has been 
> copied to what.  So if anyone wants to volunteer to answer a few (not many) 
> questions ....

I am not very knowlegeable with regard to CVS. I did learn about "cvs
diff" this morning for an unrelated project...

> Others will be able to answer this.  It would certainly be a good way of 
> testing things before a "big bang" at the Eisteddfod.


> > 	xmodmap -e 'keycode 66 = Multi_key' <etc>
> Well, I think the ability to use t?-bachs so easily on all the vowels
> except w and y is great, so this is enough for me at the minute.  I
> discussed this with Dewi, and the variability in how w-to-bach and
> y-to-bach are represented between systems led me to the compromise on
> KG/100 - use aeiou-to-bach, and w^,y^.  This is far from perfect, but
> hopefully the former will display OK on most things, and it is clear
> what the latter are.  At some point it would then be a simple matter
> to do a search-and-replace to change them for the proper code.  But
> I'm a bit vague on all these encoding things, so if anyone has
> anything better to suggest, fine.

> Thank you for your comments - my only concern at the moment is that I will 
> soon be known as the person who did NOT translate KDE :-(

Actually, I've got some more comments already, I'm afraid.

It would be nice if Omnivore displayed its results in a two-column

And I've noticed that the encodings for the web pages aren't consistent.
Omnivore appears to be UTF-8 whereas the rest of the website seems to be
using Windows 1252. Strangely (since I use Mozilla on Debian) the latter
seems to work better: letters with hetiau bychain break on the former,
although they behave again if I tell Mozilla it's in ISO 8859-14 (Latin
Celtic). I don't know why this happens or what should be done about it,
I only thought you should know.

Well, even if you don't actually do any translating of KDE yourself, I
think you can be proud of how much you helped others' translation work.


gnome-cy mailing list
gnome-cy pengwyn linux org uk

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]