Hi Doru,

By "the old version of Spotter" I meant whatever was in Pharo 3 that would
produce a list of search results with exact matches at the top when I'd hit
shift+enter and then type e.g. 'accept.'  Sorry to have said "Pharo 4" &
been unnecessarily confusing.


I'm just going to use 'accept' for this example but why show the mixed list
of 637 implementors of accept* and not lead with "accept".  Why was it
decided that inexact matches to the typed input be privileged above exact
matches in the new tool?  Is it a bad Levenshtein distance algorithm or
something?


Thanks for helping me figure it out


Paul





Tudor Girba-2 wrote
> Hi,
> 
> What do you mean by the old version of Spotter?
> 
> Just in case: you should know that Spotter is made to be extensible. This
> means that if you want to play with your own way of searching for objects,
> you can just do it. Let me know if you to try and if you need help in this
> direction.
> 
> Cheers,
> Doru
> 
> 
> 
> On Wed, Jun 10, 2015 at 9:10 PM, Paul DeBruicker <

> pdebruic@

> > wrote:
> 
>>
>> Is there any way to change back to the old version of Spotter in Pharo 4?
>>
>>
>>
>>
>>
>> Nicolai Hess wrote
>> > 2015-06-10 16:24 GMT+02:00 Paul DeBruicker <
>>
>> > pdebruic@
>>
>> > >:
>> >
>> >> So by default the search tool is only guaranteed to return an exact
>> term
>> >> match if there are only less than 5 non-exact match results?
>> >>
>> >>
>> > Yes
>> >
>> >
>> >
>> >
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> Nicolai Hess wrote
>> >> > 2015-06-10 7:39 GMT+02:00 Paul DeBruicker <
>> >>
>> >> > pdebruic@
>> >>
>> >> > >:
>> >> >
>> >> >> when I hit shift+enter and type 'accept' I get things that are not
>> >> >> #accept, e.g. #accept: and AbstractAcceptor.
>> >> >>
>> >> >> If I add a space after accept it doesn't help.
>> >> >>
>> >> >>
>> >> >> What do I not understand?
>> >> >>
>> >> >
>> >> > the result list is not sorted and the result list is built by all
>> >> methods
>> >> > having the query string as part
>> >> > of its selector name.
>> >> >
>> >> > Yes this can be improved and it is not difficult, for example you
>> can
>> >> add
>> >> > this method to
>> >> >
>> >> > GTFilterImplementor>>applyFilterWithQuery
>> >> >     super applyFilterWithQuery.
>> >> >     items sort: [ :a :b | (self itemFilterNameFor: a) size < (self
>> >> > itemFilterNameFor: b) size ]
>> >> >
>> >> > this will sort the result list by the size of the selector name. So,
>> if
>> >> > there is a perfect match,
>> >> > it will be listed first.
>> >> > (BUT only in the implementors category if you "dive-in", not in the
>> >> > 5-elements-result-preview-list).
>> >> >
>> >> > Maybe there is  a better way without sorting. (We can modify
>> >> > applyFilterWithQuery for the implementors
>> >> > filter, to put perfect matches at the begining of the list).
>> >> >
>> >> > But all this is not easy to discover. Spotter classes make some
>> heavy
>> >> use
>> >> > of delegation, many operations
>> >> > are split and delgated to subclasses (GOOD!)
>> >> > many classes aren't documented (BAD!) and this makes it really
>> >> difficult
>> >> > to
>> >> > catch how all this is supposed to work together.
>> >> >
>> >> >
>> >> > nicolai
>> >> >
>> >> >
>> >> >
>> >> >
>> >> >>
>> >> >>
>> >> >> Thanks
>> >> >>
>> >> >>
>> >> >> Paul
>> >> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> --
>> >> View this message in context:
>> >>
>> http://forum.world.st/Using-GTSpotter-how-do-I-find-an-implementor-of-accept-tp4831299p4831428.html
>> >> Sent from the Pharo Smalltalk Users mailing list archive at
>> Nabble.com.
>> >>
>> >>
>>
>>
>>
>>
>>
>> --
>> View this message in context:
>> http://forum.world.st/Using-GTSpotter-how-do-I-find-an-implementor-of-accept-tp4831299p4831506.html
>> Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com.
>>
>>
> 
> 
> -- 
> www.tudorgirba.com
> 
> "Every thing has its own flow"





--
View this message in context: 
http://forum.world.st/Using-GTSpotter-how-do-I-find-an-implementor-of-accept-tp4831299p4831566.html
Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com.

Reply via email to