Hi Christoph,


> My personal opinion is that the menu-filtering is actually more convenient
> (at least to me). If you type a non-matching filter term into a list morph
> (for example, "hello world" into the System Browser's class list), the
> filtering "breaks" half the way and a partial match is selected. If you are
> doing the same in a menu morph, however, each character is recorded so you
> can correct typing errors, and you do not need to check your results before
> pressing Enter. It happens so often that I type something wrong and get
> irritated as the whole list "stops working", not accepting any further
> character. I think menus are just convenient as they are, rather I would
> change the list implementation to show nothing (but maybe a "no-hit
> message"?) if nothing matches your filter term.
>
You're right.  And, you've even reminded me of one of my own UI principles
that, in general, a unidirectional interaction -- where the user can direct
the software without needing visual feedback -- is better wherever
possible.  In this case, you can input via keyboard without having to read
the screen, and even if you make a typing mistake, you can still direct it
to your selection unidirectionally.  [End] key probably not hit while
filtering, but when [Backspace]ing, so I think your tweak is fine.  :)

Lists aren't as predictable as menus, so two-way interaction is probably
unavoidable, and if we displayed an empty list there'd be no place to
display the current filter string..

However, we were talking about refactoring. Do you actually think that
> "keyValue >= 32" in a morph class should be preferred to a string utility
> that gives the thing we want to do a clear name?
>

No, I think clarity is better.  But not a new extension method.  More like
how you improved the line higher up in the method from

     evt keyValue = 8 " Character backspace asciiValue "

to

     evt keyCharacter = Character backspace

I think you should've stuck with your improved pattern for the >=32 check,
e.g.,

    evt keyCharacter >= Character space

and maybe a terse comment for why.  I certainly wouldn't make a whole new
version of Morphic only for that, though.

Best,
  Chris





>
> Best,
>
> Christoph
> ------------------------------
> *Von:* Squeak-dev <squeak-dev-bounces at lists.squeakfoundation.org> im
> Auftrag von Chris Muller <asqueaker at gmail.com>
> *Gesendet:* Montag, 7. Oktober 2019 20:30:42
> *An:* The general-purpose Squeak developers list
> *Betreff:* Re: [squeak-dev] #isControlCharacter (was: The Trunk:
> Morphic-mt.1514.mcz)
>
> On Mon, Oct 7, 2019 at 1:29 AM Marcel Taeumel <marcel.taeumel at hpi.de>
> wrote:
>
>> Menu filtering is so cool! :-) And convenient.
>>
>> This proposal about #isControlCharacter aims at preventing malformed user
>> input. For example, the user might wonder if a filter does strange things
>> only because the system cannot render certain (wrongly) typed characters.
>>
>> An easier solution would be to render "?" or something similar for such
>> characters. Then, the user can recognize the bad input and hit the
>> [backspace] key. This way, we do not constrain ourselves to a specific
>> range of menu labels at all. :-)
>>
>
> -1.  We should let the machine do the recognition and backspacing -- e.g.,
> any character which causes all selections of a menu to be grayed out should
> cause that character to be rejected.  Perhaps with a beep or flash of the
> menu.  Note this would provide a more consistent UI with the
> list-filtering, which behaves that way.
>
> Menus should be constrained English, with use of #translated for other
> languages with special characters.
>
> Best,
>   Chris
>
>
>
>>
>> Best,
>> Marcel
>>
>> Am 03.10.2019 02:34:40 schrieb David T. Lewis <lewis at mail.msen.com>:
>> My personal event horizon does not extend beyond seven bits, so
>> anything that I say about Unicode should not be taken seriously.
>> However:
>>
>> The concept of "isControlCharacter" dates back to the days of data
>> terminals connected over seven bit serial lines, and to the use of
>> control
>> characters for various functions not related to printing characters on
>> the display screen. I do not really understand the reference to
>> "type-to-filter a menu for things like '...' or ' ' (space)", but
>> it does not sound to me like "control character" would be the right
>> description.
>>
>> Wikipedia has http://en.wikipedia.org/wiki/Control_character which
>> gives good background. It includes a paragraph about Unicode that
>> says this:
>>
>> In Unicode, "Control-characters" are U+0000—U+001F
>> (C0 controls), U+007F (delete), and U+0080—U+009F
>> (C1 controls). Their General Category is "Cc". Formatting codes
>> are distinct, in General Category "Cf". The Cc control characters
>> have no Name in Unicode, but are given labels such as ""
>> instead.
>>
>> So it certainly would be possible to define #isControlCharacter
>> for Unicode characters, but IMHO it seems to drag along some ancient
>> concepts of serial data communications that are probably not
>> meaningful inside a Squeak image.
>>
>> Another reference is "man 3 isalpha" on Unix/Linux (C89 and Posix).
>> There are quite a few character testing functions. Most of them
>> do not exist in Squeak, and I would not be inclined to start adding
>> any of them without a good reason.
>>
>> Dave
>>
>>
>> On Wed, Oct 02, 2019 at 11:05:04PM +0000, Thiede, Christoph wrote:
>> > Hi,
>> >
>> >
>> > any opinions about this proposal? :-)
>> >
>> >
>> > Best,
>> >
>> > Christoph
>> >
>> > ________________________________
>> > Von: Squeak-dev im Auftrag von Thiede, Christoph
>> > Gesendet: Donnerstag, 5. September 2019 15:06 Uhr
>> > An: squeak-dev at lists.squeakfoundation.org;
>> packages at lists.squeakfoundation.org
>> > Betreff: Re: [squeak-dev] The Trunk: Morphic-mt.1514.mcz
>> >
>> >
>> > Personally, I sometimes type-to-filter a menu for things like '...' or
>> ' ' (space). Am I the only one who does this? :)
>> >
>> >
>> > I still would consider something like Character>>#isControlCharacter
>> useful. How about this?
>> >
>> >
>> > Unicode class>>isControlCode: charCode
>> > ^ (self generalCategoryOf: charCode) <= cs="">
>> >
>> > Character>>isControlCharacter
>> > ^ self encodedCharSet isControlCode: self charCode
>> >
>> > Versions for other char sets would have to be added.
>> >
>> > Best,
>> > Christoph
>> > ________________________________
>> > Von: Squeak-dev im Auftrag von commits at source.squeak.org
>> > Gesendet: Donnerstag, 5. September 2019 14:36:59
>> > An: squeak-dev at lists.squeakfoundation.org;
>> packages at lists.squeakfoundation.org
>> > Betreff: [squeak-dev] The Trunk: Morphic-mt.1514.mcz
>> >
>> > Marcel Taeumel uploaded a new version of Morphic to project The Trunk:
>> > http://source.squeak.org/trunk/Morphic-mt.1514.mcz
>> >
>> > ==================== Summary ====================
>> >
>> > Name: Morphic-mt.1514
>> > Author: mt
>> > Time: 5 September 2019, 2:36:47.777837 pm
>> > UUID: 49f63b40-72ce-c447-9fb0-9400ea43fab2
>> > Ancestors: Morphic-mt.1513, Morphic-ct.1501
>> >
>> > Merges Morphic-ct.1500. and 1501.
>> >
>> > Can't we just filter for #isAlphaNumeric? Do we need parentheses etc.?
>> Would be more readable than ">= 32". I also do not think that #caseOf:
>> helps much in terms of readability. ;-)
>> >
>> > =============== Diff against Morphic-mt.1513 ===============
>> >
>> > Item was changed:
>> > ----- Method: MenuMorph>>handleFiltering: (in category 'keystroke
>> helpers') -----
>> > handleFiltering: evt
>> >
>> > | matchString |
>> > matchString := self valueOfProperty: #matchString ifAbsentPut: [ String
>> new ].
>> > + matchString := true
>> > + caseOf: {
>> > + [ evt keyCharacter = Character backspace ] ->
>> > + [ matchString isEmpty
>> > + ifTrue: [ matchString ]
>> > + ifFalse: [ matchString allButLast ] ].
>> > + [ evt keyValue >= 32 ] ->
>> > + [ matchString , evt keyCharacter ] }
>> > + otherwise: [ matchString ].
>> > - matchString := evt keyValue = 8 " Character backspace asciiValue "
>> > - ifTrue: [
>> > - matchString isEmpty
>> > - ifTrue: [ matchString ]
>> > - ifFalse: [ matchString allButLast ] ]
>> > - ifFalse: [
>> > - matchString copyWith: evt keyCharacter ].
>> > self setProperty: #matchString toValue: matchString.
>> > self displayFiltered: evt!
>> >
>> >
>>
>> >
>>
>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20191025/523db1e3/attachment.html>