Auto-completion "eats" next symbol

[Sven Brauch]

Yeap, right. There's a point in how it works now, but it's just not
what you usually want to do. I'd also support changing it to adding
the word instead of replacing existing stuff.

Regards,
Sven

2010/12/9 Dan Leinir Turthra Jensen <admin at leinir.dk>:
> On Thursday 09 Dec 2010 14:41:58 Aleix Pol wrote:
>> On Thu, Dec 9, 2010 at 3:17 PM, Julien Desgats
> <julien.desgats at gmail.com>wrote:
>> > Hey,
>> > I've noticed a particularly disturbing behavior (bug?) on
>> > auto-completion : when you type something while there is characters
>> > after cursor, they are erased until the end of word if you use
>> > auto-completion.
>> >
>> > For instance if you have ("][" is the cursor)
>> >
>> >   f][bar(0);
>> >
>> > And auto-completion suggest "foo", if you validate that suggestion,
>> >
>> > you will get :
>> >   foo(][0);
>> >
>> > As you can see, the "bar" symbol has been erased. I think the correct
>> >
>> > behavior should be :
>> >   foo][bar(0);
>> >
>> > or
>> >
>> >   foo(][bar(0);
>> >
>> > But if you are in the following situation :
>> >   f][ bar(0);  // notice the space between the cursor and "bar"
>> >
>> > The result of auto-completion will be :
>> >   foo(][) bar(0);
>> >
>> > As a result, I'm constantly forced to add spaces in such situations to
>> > avoid auto-completion to eat existing symbols !
>> >
>> > So, what do you think about this idea of improvement ?
>> >
>> > Regards,
>> > Julien
>> >
>> > --
>> > KDevelop-devel mailing list
>> > KDevelop-devel at kdevelop.org
>> > https://barney.cs.uni-potsdam.de/mailman/listinfo/kdevelop-devel
>>
>> I've considered that many times. It's definitely easy to fix, I like it
>> when I can just change the current word by triggering the completion.
>>
>> What do others think?
>>
>> Aleix
>
>  Could not agree more - the auto completion is supposed to work best in the
> majority of cases, where right now it annoys in the majority of cases... While
> it does seem like a good idea on the surface to be editing the symbol you're
> on rather than adding a new one, it does seem to turn out that well, that just
> isn't the case. So yup, just as a user (not a usability guide here, this sort
> of thing would need testing we just don't really have the time for), i would
> certainly be glad to see the behaviour changed as suggested :)
>
> --
> ..Dan // Leinir..
> http://leinir.dk/
>
>                          Co-
>                            existence
>                          or no
>                            existence
>
>                          - Piet Hein
>
> --
> KDevelop-devel mailing list
> KDevelop-devel at kdevelop.org
> https://barney.cs.uni-potsdam.de/mailman/listinfo/kdevelop-devel
>