[QtExtended] some things
lpotter at trolltech.com
Mon Mar 16 04:42:01 CET 2009
Dale Maggee wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> Lorn Potter wrote:
>> Typing numbers are not 'suggested'.
> Fine, be fastidious if you must, You know what I meant.
> How long did it take you to type your postal address? Now, how long does
> it take on your PC?
> Oh... you didn't test it? You didn't bother typing out the sample SMS in
> my email? Hmmm, Interesting...
Quite actually, I was using the predictive keyboard long before anyone
on this list was, so yes, I have done this. Many times.
>> I personally do not find the delay all that troublesome.
> Well, that indicates to me that you're either not using it for everyday
> use, or that you happen to be a lucky person who's typing dictionary
> words, or, more likely, that as the developer all the words you use
> commonly happen to be in the dictionary already (strange, that...)
>> As I said, if you can easily switch to the qwerty keyboard.
> bwahahaha, that's a funny joke. Have you ever tried actually switching
> to the qwerty keyboard using just your finger? for me, it usually takes
> at least 3-4 attempts just to bring up the list of input methods. Then
> there's actually attempting to type on the qwerty keyboard, which we'll
> get into later...
umm, yes. I can access the menu just fine with my finger.
Options -> Change Input Method.
>> This is not a very valid usecase. How many times are you going to type
>> this into your phone in a real world situation?
> You're incorrect - it's a completely valid, worst-case scenario which
> serves very well to illustrate the inadequacy of the 'typing one letter
> at a time' method which you are advocating. In real-world usage, I'll
> concede that 'antidisestablishmentarianism' isn't used often, but long
> words which aren't in the dictionary are not at all uncommon, especially
> given long words combined with sms-speak spelling. How would you suggest
> a biologist or chemist use this input method to talk about his work?
A use case is usually something that is very often used and repeatable
for any user. Typing 'antidisestablishmentarianism' is hardly typical.
> If you're trying to tell me that the biologist needs to put a
> "biologist's dictionary" on his phone before he can use it, I'll laugh
> my ass off, and I'll tell you exactly how utterly ludicrous that Idea
> really is.
No, but he will need to add those words to the common dictionary before
they will show up in the list of words. To enter a word in the
dictionary - simply press and hold the letters method.
> As a software developer, I'm *very* interested in worst-case scenarios,
> as they work very well to find all these little inefficiencies - if I
> can make my software usable in a worst case scenario, it's going to be
> absolutely brilliant for average, everyday use. I test (and sometimes
> write) all my software on low-powered machines for this reason
There are too many niche scenarios to target, so we targeted common,
most used ones.
> If you're not interested in actually making your software robust and
> useful, maybe you should apply for a job at Microsoft.
>> For a fact, though, this took me exactly as much time to tap it out,
>> without having to use the press down method ~10 seconds.
> Care to clarify on that? to me, this statement makes no sense.
Using the 'predictive keyboard', it took me 10 seconds to type
> If you're saying that you just typed in 'antidisestablishmentarianism'
> and it was suggested and you were able to choose it because it had no
> idea what you were typing and showed you what you typed, then you're
> doing it wrong, and your test is invalid.
No, it was not a suggested word, as it was longer than any suggestions
it could find, so it just took the letters I was typing in.
> For it to be a valid test, you need to do the following:
> 1. *quickly* Type 'antidisestablishmentarianism', as if you're expecting
> it to be in the dictionary. (Type quickly, and ignore precision,
> assuming the dictionary will still suggest the correct word even if you
> accidentally hit a couple of letters wrong).
I did. its not in the dictionary. see my statement above.
> (i.e: If I'm typing 'owned' in an SMS, I don't concern myself if I
> actually accidentally type "pwned", because I know the dictionary will
> pick up on my mistake and suggest "owned". For your test to be valid,
> you need to type 'antidisestablishmentarianism' as if you're expecting
> it to be in the dictionary - i.e quickly and without precision)
> 2. Since you typed quickly and without precision, you'll now have
> something like "antidiseatablishmentaroanusm" (note that my spelling
> wasn't precise, and I now have an incorrectly spelled word being
> 3. A this point, If I swipe my finger to the left (i.e: backspace) to
> correct the spelling mistakes, the entire word is erased, and I have to
> do the 'one letter at a time' bullshit.
> 4. Alternatively at step 3, I could have tapped on the incorrectly
> spelled word to have it added to the sms I'm composing, and then edited
> it by positioning the cursor, backspacing, fixing the mistakes, and
> repositioning the cursor at the end of the text.
> I can do that, that is, *if* the software happens to want to let me
> position the cursor, rather than selecting everything between where I'm
> trying to put the cursor and the end of the message. I haven't been able
> to determine exactly what factors influence whether it will let me
> position the cursor or select text to the end, but I think it has to do
> with the ambient humidity and wind speed/direction.
> Of course, This is all as much or more of a hassle than the 'one letter
> at a time' method which you advocate.
Once you do that, your missEnglish word will be in the dictionary, or it
should be. Then it will find it and you won't have to tap the whole word
ever again. Thus saving you hours of tapping time if you would have used
even a desktop keyboard.
> Don't you dare even trying to tell me that my test is invalid, and that
> I need to type with precision in step 1, because if you try to tell me
> that, then I'll have one very simple and effective retort: What is the
> point of having a dictionary lookup system like the one in QTe unless
> it's to figure out what you were trying to type, and correct your
> mistakes? If I still need to type with precision, then the dictionary
> stuff is a completely useless feature which is eating up CPU cycles and
> screen real-estate for absolutely no benefit, and it might as well be
no, I did not say that, you will only need to type with precision only
once. How is any system supposed to find a word not in the dictionary
that it uses?
> Don't tell me "you can delete the dictionary file", because this doesn't
> turn off the dictionary crap, and it only makes life harder - it still
> suggests the names of contacts etc - the only difference is that now
> instead of letter-by-letter typing maybe one in every 3 words, you're
> letter-by-letter typing 97 out of every 100 words. Deleting the
> dictionary file and typing with precision to make it suggest what you
> typed is similarly useless, because if you make a spelling mistake,
> 'backspace' erases the entire word.
>> You can even use this with big fingers and hit in between letters and it
>> will still work (suggestively) well.
> That's possibly the funniest thing I've ever read!
really. wow, reality is that funny.
> I don't know what universe you've been doing your testing in, but in the
> universe I live in, I can barely use the qwerty keyboard *with a
> stylus*! Trying to use it with my fingers, well, that's just a really
> really unfunny joke - the huge delays on the predictive keyboard are
> easier and less frustrating. I'll give you $500 in cash if you can type
> the phrase "this input method is godawful" on the qwerty keyboard using
> only your fingers, while sitting on a moving train, without making any
> mistakes, in less than 5 minutes... Or are you going to try to argue
> that this isn't a valid use-case either?
No one said to use the little qwerty keyboard with fingers.
>> Don't make it something its not
>> supposed to be.
> Don't pretend your software isn't shit when it clearly and demonstrably is.
If you don't like it, either stop using it, or grab an editor and send
in a patch. and stop the personal attacks. Its very unbecoming.
> Telling users that they're imagining the problems they're having is not
> a way to solve issues.
I am merely explaining to you the way it works.
> Ooooh, yay! I get to use one of my example words:
> You just got pwned.
> After that last email, I'm glad QTe's development is no longer in your
> hands - hopefully the community will actually be interested in
> usability, quality and robustness.
What do you mean, it's always in my hands. It's in the hands of anyone
who wants it. Including you. It has been for a while now, too.
I would like to see you write phone software that is even as close to
working as well as Qtopia.
ignorance is bliss, I guess.
Software Engineer, Qt Software R & D, Nokia Pty Ltd.
More information about the community