I've spent this morning wrestling with this. Right now when a model (eg, ArraySelectionModel, ArrayListSelectionModel) has its dataProvider replaced, it will set the selectedIndex to -1 (and selectedItem to null) if the dataProvider is being set to null; otherwise it leaves the values unchanged unless the current value exceeds the size of the new dataProvider.
A typical way to change things is to update the dataProvider and then set the list.dataProvider to null (selectedIndex becomes -1, dispatches "selectedIndexChanged") followed setting list.dataProvider back to the original dataProvider. This is done because we did not have a dynamic system. Reseting the selection in this cause falls on the application writer. For dynamic data, I think we also need a dynamic model (eg, DynamicCollectionViewSelectionModel) which is tuned into the needs of a dynamic system. Before changing a dataProvider, the DyanmicCollectionViewSelectionModel will dispatch a "dataProviderChanging" event. This would give a selection restoration bead the opportunity to store the current selection. Then upon receiving the "dataProviderChanged" event, restore the selection. Unfortunately, the selection restoration bead still does not know when an item will be added or removed. While the model can listen for a complete replacement of the dataProvider and warning with a "dataProviderChanging" event, then is no "itemAdding" or "itemRemoving" event prior to an item being added or removed. To do this, we will need to having a new collection view that extends ArrayList and dispatches those extra events. Then the selection restoration bead can operate correctly. This is my current thinking. —peter On 1/2/18, 4:43 PM, "Alex Harui" <[email protected]> wrote: >And this one reason we have beads as a pattern. There is no one right way >to preserve selection and selection should not always be preserved for >certain kinds of data provider changes. So, create a bead that performs >the selection preservation you need and give it a name that makes it clear >what it does. Then others can create similar beads for other strategies. > >My 2 cents, >-Alex > >On 1/2/18, 1:26 PM, "Piotr Zarzycki" <[email protected]> wrote: > >>Peter, >> >>You didn't misread, but my thoughts probably didn't far away into problem >>resolution. That's why I was surprise why I haven't thought about >>preserving selection. :) >> >>Thanks, Piotr >> >> >>2018-01-02 22:10 GMT+01:00 Peter Ent <[email protected]>: >> >>> I thought because you mentioned that rollover and selection were messed >>> up, that was the thing needing attention. I must have misread your >>>email; >>> I'll go back. >>> >>> In the meantime, I have a Button that inserts an item into the >>> dataProvider. Before doing this I am: >>> >>> var oldSelection:Object = list.selectedItem; // or selectedIndex?? >>> >>> >>> then update the dataProvider by inserting an item into it. For a List >>>with >>> the standard bead set, I have to: >>> >>> list.dataProvider = null; >>> list.dataProvider = applicationModel.dataProvider; >>> >>> After that, I just restore the selection: list.dataProvider = >>>oldSelection >>> >>> Need to figure out if the dataProvider will change, save the selection, >>> allow the change, then restore it. >>> >>> —peter >>> >>> On 1/2/18, 3:58 PM, "Piotr Zarzycki" <[email protected]> wrote: >>> >>> >How are you doing that preservation, by simply saving item somewhere >>> >before >>> >remove/add operation ? >>> > >>> >I'm not following the part where we you wanted some opinion. Actually >>>in >>> >case of refreshing I didn't even thing about saving selection, but >>>it's >>> >true that it should happen! It's surprises me that in my app I forgot >>> >about >>> >that. :) >>> > >>> >Thanks, Piotr >>> > >>> > >>> >2018-01-02 21:44 GMT+01:00 Peter Ent <[email protected]>: >>> > >>> >> The first task is to determine if a list's selectedIndex or >>>selectedItem >>> >> should be preserved. >>> >> >>> >> For selectedIndex, adding or removing an item will get the same row >>> >> selected until that row no longer exists. >>> >> >>> >> For selectedItem, adding or removing an item will keep the same item >>> >> selected until that item no longer exists. For example, if the 4th >>>item >>> >>is >>> >> "red" and you insert an item at position 2, "red" remains selected >>>by >>> >> shifts down a row. >>> >> >>> >> I can see this working both ways. >>> >> >>> >> In either case, first preserving the selection (index or item) >>>before >>> >>the >>> >> add or remove and then restoring it works fine - I did this in >>> >>application >>> >> code, not in framework code. Works very nicely, as a matter of fact. >>> >>Works >>> >> for both traditional lists and dynamic lists with the new beads >>>(yours, >>> >> Piotr, or my new collection ones). Its *very simple* code. >>> >> >>> >> So once we decide which selection is preserved, we need to decide >>>where >>> >> that goes. Even though this code is just a couple of lines, in the >>>PAYG >>> >> world, not every app needs this so it must go into additional beads >>> >>since >>> >> it is added functionality. >>> >> >>> >> —peter >>> >> >>> >> >>> >> >>> >> >>> >> >>> >> On 1/2/18, 1:31 PM, "Piotr Zarzycki" <[email protected]> >>>wrote: >>> >> >>> >> >Hi Peter, >>> >> > >>> >> >Welcome after vacation! I sense all ideas when I'm working with >>>that >>> >> >stuff. >>> >> >I think you will see the issue with indexes, even if it's not the >>>case >>> >> >with >>> >> >removing from the list. I will wait for your review with that index >>> >> >refresh. >>> >> > >>> >> >Thank you so much! :) >>> >> > >>> >> > >>> >> >2018-01-02 19:10 GMT+01:00 Peter Ent <[email protected]>: >>> >> > >>> >> >> Hi, >>> >> >> >>> >> >> Catching up and building stuff. I'm going to repeat some things >>>for >>> >>the >>> >> >> befit of anyone joining Royale for 2018 (and to refresh my own >>> >>thoughts >>> >> >> coming off of holiday!). >>> >> >> >>> >> >> The initial idea of List and related classes was their beads were >>>to >>> >>be >>> >> >> almost static. You'd have data that you would use to create the >>> >> >> dataProvider to the list and when set, would cause the >>>itemRenderers >>> >>to >>> >> >>be >>> >> >> deleted and replaced. Even something as simple as adding new data >>>to >>> >>the >>> >> >> end of the dataProvider would require this reset: you'd have to >>>clear >>> >> >>the >>> >> >> dataProvider (list.dataProver = null) and set it again or create >>>a >>> >> >> completely new dataProvider and set that to the list. >>> >> >> >>> >> >> The selected item and index should, in this case, be cleared. I'd >>> >>have >>> >> >>to >>> >> >> run a test to see if that really happens as I don't remember. If >>>you >>> >> >>want >>> >> >> to reset them back (assuming that index still existed) you would >>> >>write a >>> >> >> bead to do it. >>> >> >> >>> >> >> Now we are at the dynamic stage where we want to handle runtime >>> >>updates >>> >> >>to >>> >> >> the dataProvider and create just new itemRenderers and insert >>>them or >>> >> >> remove itemRenderers when data is deleted from the dataProvider. >>> >> >> >>> >> >> Alex believes that the initial creation and insertions are the >>>most >>> >> >>common >>> >> >> of the dynamic actions; removals being less common. As such in >>>the >>> >>PAYG >>> >> >> world you would have a bead that handled a replacement of the >>> >> >>dataProvider >>> >> >> (which covers the initialization step) and a bead which handled >>> >> >>additions >>> >> >> and a bead to handle removals. There's a lot of duplication of >>>code >>> >>in >>> >> >> these and under a non-PAYG system you'd just combine these into a >>> >>single >>> >> >> bead to do initialize, add, and removal. >>> >> >> >>> >> >> In the dynamic beads, the selection should be preserved. That >>> >> >>information >>> >> >> is stored in the model. The model doesn't do much else accept >>>keep >>> >>the >>> >> >> data. The standard view bead would not know about these extra >>>events >>> >>so >>> >> >> yes, we'd need another view bead to intercept those events and >>> >>re-apply >>> >> >>or >>> >> >> adjust the selection. >>> >> >> >>> >> >> ‹peter >>> >> >> >>> >> >> On 12/30/17, 4:29 PM, "Piotr Zarzycki" >>><[email protected]> >>> >> >>wrote: >>> >> >> >>> >> >> >Peter, >>> >> >> > >>> >> >> >I have figure out what is wrong in case of List. I reported in >>>the >>> >> >> >previous >>> >> >> >email that in the ListExample once you remove one item, >>>something is >>> >> >>wrong >>> >> >> >with selection. Issue happened when you removing item in the >>>middle >>> >>of >>> >> >> >collection. >>> >> >> > >>> >> >> >Let's say that you have 5 items and you are removing item with >>> >>index = >>> >> >>2. >>> >> >> >Removing will work properly, but StringItemRenderer -> >>> >> >> >DataItemRenderer -> UIItemRendererBase >>> >> >> ><- has property "index" which is basically not being refreshed >>>after >>> >> >> >remove. In the result renderers has following indexes: >>> >> >> > >>> >> >> >0, >>> >> >> >1, >>> >> >> >3, >>> >> >> >4 >>> >> >> > >>> >> >> >Which is messes up selection and roll over color changes. How >>> >>actually >>> >> >>fix >>> >> >> >that ? My first thought was that ListView should listen for >>> >> >>"layoutNeeded" >>> >> >> >(we are dispatching that event after remove, add etc.) and >>>refresh >>> >> >>indexes >>> >> >> >of all renderers. >>> >> >> > >>> >> >> >The second thought is that ListView should listen for remove >>>event, >>> >> >>where >>> >> >> >we will have index of removed row, so we will be able loop >>>through >>> >> >>part of >>> >> >> >the collection and refresh rest of the indexes. It is more >>> >>efficient, >>> >> >>but >>> >> >> >maybe this is job for new type of View (bead) - ListRemoveView >>>bead >>> >>? >>> >> >> > >>> >> >> >Looking forward to your thought once you get back from Vacation. >>>:) >>> >> >> > >>> >> >> >Thanks, Piotr >>> >> >> > >>> >> >> > >>> >> >> > >>> >> >> >2017-12-22 15:05 GMT+01:00 Piotr Zarzycki >>> >><[email protected]>: >>> >> >> > >>> >> >> >> Hi Peter, >>> >> >> >> >>> >> >> >> I haven't checked your branch yet, but I just pushed couple of >>> >> >>things to >>> >> >> >> the develop which I needed for my examples. Here is the list >>>of >>> >>beads >>> >> >> >>which >>> >> >> >> we currently have and can be used for List. I think names are >>>self >>> >> >> >> explanatory. >>> >> >> >> >>> >> >> >> DynamicAddItemRendererForArrayListData >>> >> >> >> >>> >> >> >> DynamicUpdateItemRendererForArrayListData >>> >> >> >> >>> >> >> >> DynamicRemoveItemRendererForArrayListData >>> >> >> >> >>> >> >> >> DynamicRemoveAllItemRendererForArrayListData >>> >> >> >> >>> >> >> >> DynamicRemoveAllByNullItemRendererForArrayListData >>> >> >> >> >>> >> >> >> All beads above should be used once you set for your list: >>> >> >> >>IDataProviderItemRendererMapper: >>> >> >> >>ClassReference("org.apache.royale.html.beads. >>> >> >> DataItemRendererFactoryForAr >>> >> >> >>rayList") >>> >> >> >> >>> >> >> >> We have also DynamicItemsRendererFactoryForArrayListData - >>>which >>> >>is >>> >> >> >>doing actually two things. Create all items once someone set >>> >> >> >>dataProvider (the same logic as in >>> >> >>DataItemRendererFactoryForArrayList) >>> >> >> >>AND handle adding to >>> >> >> >> >>> >> >> >> ArrayList (the same logic as in >>> >> >>DynamicAddItemRendererForArrayListData) >>> >> >> >>- It duplicates logic from those both beads. >>> >> >> >> >>> >> >> >> >>> >> >> >> Two Scenarios with >>>DynamicItemsRendererFactoryForArrayListData: >>> >> >> >> >>> >> >> >> 1) If we use DynamicItemsRendererFactoryForArrayListData >>>simply >>> in >>> >> >>the >>> >> >> >>List as it is in our ListExample [1] we don't have to remember >>>to >>> >> >>setup: >>> >> >> >> >>> >> >> >> IDataProviderItemRendererMapper: >>> >> >> >>ClassReference("org.apache.royale.html.beads. >>> >> >> DataItemRendererFactoryForAr >>> >> >> >>rayList"). >>> >> >> >> >>> >> >> >> However in that scenario we will have in List totally useless >>>bead >>> >> >> >>DataItemRendererFactoryForArray which is being added default to >>> the >>> >> >> List. >>> >> >> >> >>> >> >> >> 2) If we use DynamicItemsRendererFactoryForArrayListData and >>>we >>> >>add >>> >> >>in >>> >> >> >>CSS IDataProviderItemRendererMapper: >>> >> >> >>ClassReference("org.apache.royale.html.beads. >>> >> >> DataItemRendererFactoryForAr >>> >> >> >>rayList") >>> >> >> >> >>> >> >> >> I'm not sure what can happen actually, cause both beads are >>>using >>> >>the >>> >> >> >>same logic to create item renderers. Can be situation when one >>> >>bead is >>> >> >> >>create renderers and second one re do this work again ? :) >>> >> >> >> >>> >> >> >> >>> >> >> >> Thanks, Piotr >>> >> >> >> >>> >> >> >> >>> >> >> >> >>> >> >> >> [1] >>> >> >> >>https://na01.safelinks.protection.outlook.com/?url= >>> >> >> https%3A%2F%2Fgithub.c >>> >> >> >>om%2Fapache%2Froyale-asjs%2Fblob%2F24b09c787fc1da2d0a19895a2b2d >>> >> >> 4a440e26c5 >>> >> >> >>97%2Fexamples%2Froyale%2FListExample%2Fsrc%2Fmain% >>> >> >> 2Froyale%2FMyInitialVie >>> >> >> >>w.mxml%23L69&data=02%7C01%7Cpent%40adobe.com% >>> >> >> 7C94d879f8f6054e10c75508d54f >>> >> >> >>cc6fe8%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0% >>> >> >> 7C636502661827506659&sd >>> >> >> >>ata=W4Ih5Q0IGhGNvOX9jrmn580Qw4ljo%2BavjqABV4%2BJr%2Bc%3D& >>> reserved=0 >>> >> >> >> >>> >> >> >> >>> >> >> >> >>> >> >> >> 2017-12-19 20:35 GMT+01:00 Olaf Krueger >>><[email protected]>: >>> >> >> >> >>> >> >> >>> Peter, great to read something about these "Collections" and >>> >> >> >>>"Hierarchical >>> >> >> >>> data" stuff here! >>> >> >> >>> Thanks for working on this, have a nice time and enjoy your >>> >>break! >>> >> >> >>> >>> >> >> >>> Olaf >>> >> >> >>> >>> >> >> >>> >>> >> >> >>> >>> >> >> >>> -- >>> >> >> >>> Sent from: >>> >> >> >>>https://na01.safelinks.protection.outlook.com/?url= >>> >> >> http%3A%2F%2Fapache-r >>> >> >> >>>oyale-development.20373.n8.nabble.com%2F&data=02%7C01%7Cpent% >>> >> >> 40adobe.com >>> >> >> >>>%7C94d879f8f6054e10c75508d54fcc6fe8% >>> 7Cfa7b1b5a7b34438794aed2c178de >>> >> >> cee1%7 >>> >> >> >>>>>>C0%7C0%7C636502661827506659&sdata=Zb1nVaD5dUx1Y89Q1jj4LxqBhtlxoZ >>> >> >> FlwDFJAq >>> >> >> >>>ruTsY%3D&reserved=0 >>> >> >> >>> >>> >> >> >> >>> >> >> >> >>> >> >> >> >>> >> >> >> -- >>> >> >> >> >>> >> >> >> Piotr Zarzycki >>> >> >> >> >>> >> >> >> Patreon: >>> >> >> >>*https://na01.safelinks.protection.outlook.com/?url= >>> >> >> https%3A%2F%2Fwww.pat >>> >> >> >>reon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com >>> >> >> %7C94d879f8f6054e >>> >> >> >>10c75508d54fcc6fe8%7Cfa7b1b5a7b34438794aed2c178de >>> >> >> cee1%7C0%7C0%7C636502661 >>> >> >> >>827506659&sdata=sjS%2Fmydo1M%2FGkkREJsc5CRyA1uFXRUK76zEHFeg >>> >> >> FXtA%3D&reserv >>> >> >> >>ed=0 >>> >> >> >> >>> >> >> >><https://na01.safelinks.protection.outlook.com/?url= >>> >> >> https%3A%2F%2Fwww.pat >>> >> >> >>reon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com >>> >> >> %7C94d879f8f6054e >>> >> >> >>10c75508d54fcc6fe8%7Cfa7b1b5a7b34438794aed2c178de >>> >> >> cee1%7C0%7C0%7C636502661 >>> >> >> >>827506659&sdata=sjS%2Fmydo1M%2FGkkREJsc5CRyA1uFXRUK76zEHFeg >>> >> >> FXtA%3D&reserv >>> >> >> >>ed=0>* >>> >> >> >> >>> >> >> > >>> >> >> > >>> >> >> > >>> >> >> >-- >>> >> >> > >>> >> >> >Piotr Zarzycki >>> >> >> > >>> >> >> >Patreon: >>> >> >> >*https://na01.safelinks.protection.outlook.com/?url= >>> >> >> https%3A%2F%2Fwww.patr >>> >> >> >eon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com% >>> >> >> 7C94d879f8f6054e10 >>> >> >> >c75508d54fcc6fe8%7Cfa7b1b5a7b34438794aed2c178de >>> >> >> cee1%7C0%7C0%7C636502661827 >>> >> >> >506659&sdata=sjS%2Fmydo1M%2FGkkREJsc5CRyA1uFXRUK76zEHFeg >>> >> >> FXtA%3D&reserved=0 >>> >> >> ><https://na01.safelinks.protection.outlook.com/?url= >>> >> >> https%3A%2F%2Fwww.patr >>> >> >> >eon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com% >>> >> >> 7C94d879f8f6054e10 >>> >> >> >c75508d54fcc6fe8%7Cfa7b1b5a7b34438794aed2c178de >>> >> >> cee1%7C0%7C0%7C636502661827 >>> >> >> >506659&sdata=sjS%2Fmydo1M%2FGkkREJsc5CRyA1uFXRUK76zEHFeg >>> >> >> FXtA%3D&reserved=0 >>> >> >> >>* >>> >> >> >>> >> >> >>> >> > >>> >> > >>> >> >-- >>> >> > >>> >> >Piotr Zarzycki >>> >> > >>> >> >Patreon: >>> >> >*https://na01.safelinks.protection.outlook.com/?url= >>> >> https%3A%2F%2Fwww.patr >>> >> >eon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com% >>> >> 7C96bb80bb6fef4b94 >>> >> >089008d5520f1a9f%7Cfa7b1b5a7b34438794aed2c178de >>> >> cee1%7C0%7C0%7C636505147194 >>> >> >>> >>>474371&sdata=D3%2FGi2wtNJLEO716RK18TbD1VMFdB3V >>> Kv3Br5KkZfFI%3D&reserved=0 >>> >> ><https://na01.safelinks.protection.outlook.com/?url= >>> >> https%3A%2F%2Fwww.patr >>> >> >eon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com% >>> >> 7C96bb80bb6fef4b94 >>> >> >089008d5520f1a9f%7Cfa7b1b5a7b34438794aed2c178de >>> >> cee1%7C0%7C0%7C636505147194 >>> >> >474371&sdata=D3%2FGi2wtNJLEO716RK18TbD1VMFdB3V >>> >> Kv3Br5KkZfFI%3D&reserved=0>* >>> >> >>> >> >>> > >>> > >>> >-- >>> > >>> >Piotr Zarzycki >>> > >>> >Patreon: >>> >*https://na01.safelinks.protection.outlook.com/?url= >>> https%3A%2F%2Fwww.patr >>> >eon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com% >>> 7C61ec983f6a9747bd >>> >cd7d08d552239493%7Cfa7b1b5a7b34438794aed2c178de >>> cee1%7C0%7C0%7C636505235128 >>> >887083&sdata=Y%2BHjYn%2BGj%2Bhb0VVVy7fSW%2BGEvQo2% >>> 2FUjH2EhucTbxR8Q%3D&rese >>> >rved=0 >>> ><https://na01.safelinks.protection.outlook.com/?url= >>> https%3A%2F%2Fwww.patr >>> >eon.com%2Fpiotrzarzycki&data=02%7C01%7Cpent%40adobe.com% >>> 7C61ec983f6a9747bd >>> >cd7d08d552239493%7Cfa7b1b5a7b34438794aed2c178de >>> cee1%7C0%7C0%7C636505235128 >>> >887083&sdata=Y%2BHjYn%2BGj%2Bhb0VVVy7fSW%2BGEvQo2% >>> 2FUjH2EhucTbxR8Q%3D&rese >>> >rved=0>* >>> >>> >> >> >>-- >> >>Piotr Zarzycki >> >>Patreon: >>*https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.pat >>r >>eon.com%2Fpiotrzarzycki&data=02%7C01%7Caharui%40adobe.com%7C4ed02057c74e4 >>5 >>0da30908d552277e71%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636505251 >>9 >>40120787&sdata=aJzPBX5fyLQVXr21gO9aLOIIQhVMgf3QaT4qEcsnWE4%3D&reserved=0 >><https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.pat >>r >>eon.com%2Fpiotrzarzycki&data=02%7C01%7Caharui%40adobe.com%7C4ed02057c74e4 >>5 >>0da30908d552277e71%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636505251 >>9 >>40120787&sdata=aJzPBX5fyLQVXr21gO9aLOIIQhVMgf3QaT4qEcsnWE4%3D&reserved=0> >>* >
