The Inbox: Tools-cmm.927.mcz
Locked
 
|
Chris Muller uploaded a new version of Tools to project The Inbox:
http://source.squeak.org/inbox/Tools-cmm.927.mcz ==================== Summary ==================== Name: Tools-cmm.927 Author: cmm Time: 17 December 2019, 4:14:11.72576 pm UUID: 02595c45-7a56-4481-9038-3bca791d316e Ancestors: Tools-mt.925 - Fix debugging of the initialization of Dictionary's. - Fix the order of entries rendered by inspectors of OrderedDictionary's. =============== Diff against Tools-mt.925 =============== Item was changed: ----- Method: DictionaryInspector>>calculateKeyArray (in category 'selecting') ----- calculateKeyArray "Recalculate the KeyArray from the object being inspected" + keyArray := object isInitialized + ifTrue: [ object keysInOrder ] + ifFalse: [ Array empty ]. + selectionIndex := 0! - - keyArray := object keysSortedSafely asArray. - selectionIndex := 0. - ! Item was added: + ----- Method: HashedCollection>>isInitialized (in category '*Tools-Debugger') ----- + isInitialized + ^ array notNil and: [ tally notNil ]! |
Re: The Inbox: Tools-cmm.927.mcz
|
|
This one removes the send to #asArray, it wasn't necessary. On Tue, Dec 17, 2019 at 4:14 PM <[hidden email]> wrote: Chris Muller uploaded a new version of Tools to project The Inbox: |
Re: The Inbox: Tools-cmm.927.mcz
|
Hi Chris, I don't think that #isInitialized is necessary. Under which circumstances is a Dictionary or Set not initialized? Best, Marcel
|
Re: The Inbox: Tools-cmm.927.mcz
|
|
Hi Marcel, Chris was wanting to step into "constructor" Dictionary new: 4 Le mer. 18 déc. 2019 à 10:02, Marcel Taeumel <[hidden email]> a écrit :
|
Re: The Inbox: Tools-cmm.927.mcz
|
|
Hmm... #isInitialized feels like an anti pattern that violates the basic "information hiding" principle... Could we rename it to make the special tool use more clear? Maybe #canBeUsedInInspector ? Best, Marcel
|
Re: The Inbox: Tools-cmm.927.mcz
|
|
Why not just use #ifError: in #calculateKeyArray? Shouldn't inspectors/debuggers be robust in any way? Ideally, we could use a similar solution to #errorPrintString and display a short red note if the "keys could not be displayed." Imho, this would also a more accurate representation than faking an empty list when there is actually no list at all.
Best, Christoph Von: Squeak-dev <[hidden email]> im Auftrag von Taeumel, Marcel
Gesendet: Donnerstag, 19. Dezember 2019 13:50:39 An: John Pfersich via Squeak-dev Betreff: Re: [squeak-dev] The Inbox: Tools-cmm.927.mcz
Hmm... #isInitialized feels like an anti pattern that violates the basic "information hiding" principle... Could we rename it to make the special tool use more clear? Maybe #canBeUsedInInspector ?
Best,
Marcel
Carpe Squeak!
|
Re: The Inbox: Tools-cmm.927.mcz
|
|
> Why not just use #ifError: in #calculateKeyArray? Shouldn't inspectors/debuggers be robust in any way? +1 That's the way to go in debugging tools that handle fragile objects. See ObjectExplorerWrapper >> #asString. Best, Marcel
|
Re: The Inbox: Tools-cmm.927.mcz
|
|
On Fri, Dec 20, 2019 at 6:36 AM Marcel Taeumel <[hidden email]> wrote:
I agree ONLY because of your qualifier:
:) Otherwise, I would consider it a bad idea... I'll submit another. - Chris
|
Re: The Inbox: Tools-cmm.927.mcz
|
|
> :) Otherwise, I would consider it a bad idea... Yeah, that on-error-do stuff should only be used carefully and rarely. I think there are other cases in our tools where such error handling is employed. Good for robustness as long as we can come up with a fallback like your empty array. Note that this is not the same is an #ifNil-check :-) Which I consider a bad idea also in tool land... :-) Best, Marcel
|
|
|
There's an old saying familiar to us Ancient Ones - "don't handle errors you can't actually handle".
One of the great classics of the genre was the Windows error that used to happen so frequently it was beyond a joke; the "attempt to execute an illegal instruction" error window had the hilarious option to just continue. > On 2019-12-22, at 6:22 AM, Marcel Taeumel <[hidden email]> wrote: > > > :) Otherwise, I would consider it a bad idea... > > Yeah, that on-error-do stuff should only be used carefully and rarely. I think there are other cases in our tools where such error handling is employed. Good for robustness as long as we can come up with a fallback like your empty array. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Cloister: a pretentious clam |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |