Re: [gnome-cy] Status update (looking good)

[Dafydd Harries <daf parnassus ath cx>]

On Wed, May 21, 2003 at 03:55:09PM +0100, Telsa Gwynne wrote:
> On Wed, May 21, 2003 at 02:40:13PM +0100 or thereabouts, Kevin Donnelly wrote:
> 
> > [anti-aliasing]
> > llyfnu ffontiau?

Well, it's not only used for font smoothing - the GNOME canvas library
can anti-alias all drawing done using it. However, "llyfnu" or something
similar might be what we're looking for. After much head scratching, the
best I could come up with was "datamgennu".

> > mnemonic key: cof-fysell, cof-allwedd?
> > accelerator: cyflymydd
> 
> I leave these in for the better-qualified to comment.

I think they're pretty sound.

> > Omnivore now contains over 32,000 strings.  
> 
> Whoa.

That is a lot!

> > At the AFFS conference (see below) the possibility of searching 
> > this to find identical strings to be put into the suggestion 
> > field of new tables (files) was raised - ie a sort of fuzzy 
> > matching - which was why I was interested in what Alan had done.  
> > I remain to be convinced about this - in KDE you will get things 
> > like Directory:, Directory, Directory %1, Name=Directory, and so 
> > on, and having to go into each of these to correct one character 
> > is hardly quicker than typing them in from scratch.  But if 
> > anyone thinks it would be worth experimenting with this, let me know.
> 
> It sounds rather like what you get after taking an old complete
> po file of translations, a much newer and longer list of strings
> to translate for the same program, and running msgmerge on them.
> You end up with a new merged file containing something like 
> "104 translated, 57 fuzzy and 18 untranslated". And you go 
> through the new file looking for the word 'fuzzy'. The merging
> program makes some great guesses when strings are identical,
> but not so great when they are not. So for example, I found
> this in gnome-mime-data after running it:
> 
> #, fuzzy
> msgid "Nullsoft video"
> msgstr "Fideo Microsoft"
> 
> Again, someone needs to go through, but it is certainly faster
> to delete the lines saying "#, fuzzy" and correct things like
> the above than it is to do the lot. 20 deletions of "fuzzy" and
> 37 corrections is still better than 57 from scratch.

I think this is something to be considered. I'd imagine that fuzzy
strings would show up differently in the interface, with ways of marking
strings as no longer fuzzy.

On the other hand, it might complicate matters too much and make it
less accessible. Perhaps it could be done so that suggestions are
inserted automatically, marked as fuzzy in the database, and unmarked
when it is modified. That only leaves the question of what is done to
fuzzies which are correct.

-- 
Dafydd

_______________________________________________
gnome-cy mailing list
gnome-cy pengwyn linux org uk
http://pengwyn.linux.org.uk/mailman/listinfo/gnome-cy