[kde-edu]: Parley (KVocTrain) - split translations

[Markus Büchele]

Am Mittwoch 10 Oktober 2007 14:15:48 schrieb Frederik Gladhorn:
> Hi people,
> after preventing me from feature reduction the last time, let's see, if we
> can play the same game again :)
>
> 1.) I have stumbled over another thing that seems uncomfortable to me:
> split solutions
> a) it makes my life harder
> b) it is badly implemented
> c) I don't know if it's usefull
> d) I don't like it
> e) ...
>
> Now my question is, does anyone use it?

I use it and I think it helps to know the number of answers. I'd prefer it to 
stay, but it should evaluate the different meanings in any order correctly.


> There will be changes in how an answer is corrected (yay!). 
Good because the old implementation with split solutions was buggy!


> I don't want to 
> promis anything on the short term, but it looks promising :)
> We won't ever achieve intelligent validation of answers, because computers
> and languages don't fit. But we can pretend ;)
>
> Reading the book of all books (The KVocTrain Handbook) of course enlightens
> me... except that the feature is not documented.
> Wow. What is a field? How do I know if I want to split my translation? Does
> it hurt?
>
> Now I went to the ordeal of creating a new document in KVocTrain to
> specifically test this nifty feature. And the result is...
> It looks ugly as hell.
> The "fields" are lineedits below each other.
>
> I am thinking about taking this out completely.
> My current idea is to use the great new (far from perfect) correction stuff
> and make it even greater by looking at individual words. So if you need to
> enter a sentence, individual words will appear in red and green.

Can't this be used even though you keep the spit answers option?

> Opinions and suggestions?
>
>
> 2.) Suggestion lists
> After such a long time I noticed if this feature was enabled, Parley kept
> on crashing... This was due to changes in KComboBox using QLineEdit
> internally instead of the K version and obviously overlooked in the Qt4
> porting effort because it uses an evil cast. This is still giving me
> nausea.
> Apart from doing other evil things it uglifies some parts of code pretty
> well an needs quite a few strange work arounds.
>
> So again, please comment on the usefulness of this thingy. Maybe there
> could be something like this, but improved usability wise?
> To me it sounds like cheating anyway. Use multiple choice, I'd say ;)

I had the impression that suggestion lists are there to add false friends and 
other tricky answers for the multiple choice test. If this is the case, I 
believe it is a good feature and should be kept. Not that I have ever used 
it ;-)

>
> Have a great day,
> Frederik (with his favorite code chopping screeching evil chainsaw in hand
> already)
>
>
> PS: if any of you gets around to testing sound and image, pleas let me know
> if it works at all, if it's usable, what could be made better etc...