Rollo62

Thanks very much for the efford, looks good to me. Search seems to do well in full text, at first sight, great.

But you can simply add the account data, Google still has a large free tier for free apps. If you use the map a lot, then this might get exhausted soon, and the cost may explode. For a seldom map requests the tfree tier might be OK. I also think that Goggle is not clever giving up so many interesting free projects in the past. So a lot of developers will have to look after more reliable sources then, and this could affect the whole Google Cloud business. I you don't trust once, you don't trust ever ...

Im afraid all GIS solutions are a little expensive, like TatukGIS. Probably this CartoVCL is something for you, but I don't know much about it.

Yes, thanks. Seems I missed this interesting post before. I didn't expect that much influence, also very intersting ist the LVCL project from Arnoud. Thanks god I'm not pushing towards smallest memory footprint on VCL, but maybe a LFMX project will be interesting one day

What do you look after for GIS, something like OpenStreetMap mapping, or also ReverseGeoCoding ? Maybe this little unit helps for OpenStreetMap, at least a somewhat. Not sure if this is still current, since its already a little older. But If you are able to spend a few bucks I would recommend the TMS OSM component, or the one for Google Maps. I have seen at least the Google Maps version working very nicely.

If you want to remove advanced RTTI from RTL/VCL, you have to recompile RTL/VCL, which is what OP wants. @Dalija Interesting. How much MB would we gain from removing Rtti in system libs ? I've never considered to remove anything from there, as I think its probably not worth to consider memory space in 21th century. Maybe this is another story with mobile devices still. Anyway, I use Rtti, so maybe I have to consider this in my next project.

I see another advantage in the 1. option, which is debugging: You can step/break into both cases, instead or only one.

Not 100% sure if there is some unclean code somewhere, but if you don't bind Rtti-Units it won't be linked to the project. Since I have to use Rtti I'm not aware of any wrong linking, please keep me updated where this happens exactly.

Yes, TreeView would be a natural choice for tree-like structures. But in FMX the TListView is especially recommended to be used on mobile, and it is therefor probably better supported/tested than the other, more desktop-related components. The difference could be visible maybe if a lot of items should be loaded, not sure if TreeView is performace-wise at same level with ListView. So if TreeView works well for you, its perfect, I would always have a look at ListView first, before moving anywhere else.

Is this a possible QC entry, as DocWiki is part of IDE help file ?

Goto: Where no man has gone before

Maybe this is helpful, http://docwiki.embarcadero.com/Libraries/XE7/en/System.SysUtils.TStringHelper.TrimEnd

There had been Rumors around that new MacMinis wouldnt be IntelBased, but Arm64. Havent checked that recently, but if so it might cause issues with FMX in General. I would check compatibility before if I consider Mac Mini.

I had issues with Bluetooth LE on MacBook and iMac at High Sierra, so I downgraded to Sierra do get BLE connection again. It was definitifely not a hardware issue, and also not an RadStudio issue, but I'm not sure either what went wrong. There were problems even with the Apple mouse to get connected after the update. I have read about many different issues with High Sierra at that time in Apple forums (and now probably solved by Mojave), but I'm always a little reluctant with upgrading a running system too fast. Edit: Not sure how long I could postpone upgrade still (March 2019), but after upgrade I would need 100% Bluetooth support too, otherwise I will get into some big trouble. Not sure if any changes have been in the CoreBluetooth from Sierra, to High Sierre, to Mojave. I'm afraid I have to do some more compatibility tests soon.

Here is a lot of stuff included, from Luxophia, you know that ? Not sure about bumpmaps, but its worth to have a look.

I've made some tests a while ago, and put them into an empty, commented region (for later unbury, if I need them). Maybe it helps somehow. class function TJson_Marshal_Base<T>.Internal_FromObject(AValue: TJSONObject ) : TJson_Marshal_Base<T>; var LMar: TJSONUnMarshal; begin Result := nil; if not Assigned( AValue ) then Exit; LMar := TJSONUnMarshal.Create(); // This sets DateTimeIsUTC internally by default try LMar.DateTimeIsUTC := True; // !! Important to ensure same Coding/Decoding {$REGION 'OptionalReverter'} // LMar.RegisterReverter(TToolPanel_Parcel, //TODO remove, since it never gets here // function (Data: TListOfStrings) : TObject // var // LObj : TToolPanel_Parcel; // begin // // if Length(Data) >= 9 then // begin // LObj := TToolPanel_Parcel.Create; // // LObj.FValueType := StrToIntDef(Data[0], -1); // LObj.FValueNum := Data[1]; // LObj.FValueUnit := Data[2]; // LObj.FTimestamp := StrToDateDef( Data[3], 0); // LObj.FTimestampColor := StrToIntDef( Data[4], Integer( TAlphaColorRec.Red )); // LObj.FLine1 := Data[5]; // LObj.FLine2 := Data[6]; // // LObj.FCoord.FromString_Location( Data[7] ); // // LObj.FGeoCode := Data[8]; // // if Length(Data) >= 10 then // LObj.FMultiLine := Data[9] // else // LObj.FMultiLine := ''; // // Result := LObj; // end // else // begin // Result := nil; // end; // // end // // ); {$ENDREGION} //UnMarshal will attempt to create a TTestObject from the TJSONObject data //using RTTI lookup - for that to function, the type MUST be defined in a unit Result := TJson_Marshal_Base<T>.Create; //( AData ); Result := LMar.CreateObject(TJson_Marshal_Base<T>, AValue, Result ) as TJson_Marshal_Base<T>; finally LMar.Free; end; end;

If so many people urgently (mis)use it that way, this should be a sign that there is a strong need 🙂

Hi folks, I'm trying to use TWebBrowser (embedded in the app) to show a special page in my mobile apps. These page is able to download images and other stuff from that page. When I use normal web browser such documents land in the downloads folder. Of coarse mobile apps life in a sandbox, and it seems that downloads don't land in public downloads folder. iOS seems to open the image in a separate page, which Android seems to do nothing. From the behaviour so far I can say that the embedded browsers behaves different than the default browser of the system. My questions are howto handle downloads via embedded TWebBrowser in Android and iOS ? Are downloads blocked completely on mobile ? Are downloads stored somewhere in the sandbox ? Can I configure somewhat in TWebBrowser (below surface), to finetune behaviour and where documents will land ? Can I hook into the click/ajax events of the TWebBrowser page (if I have no control over the page itself) ? Maybe a concept of control would be to use a local HTML5 wrapper page with frames around the target site, so to be able to get control about the inner events of the target site ? I would prefer to use the internal TWebBrowser, if possible, but also 3rd party solutions would be welcome Hope somebody of the mobile experts have found more insights and a good solution already. I cannot change the HTML5 page, by the way, to add something like a JS-bridge to Delphi directly, so I hope to find some hooks and tweaks around. Rollo

Thanks, I still checking some stuff to get around this at all, with Indy and THttpClient, so far I prefer the new THttpClient. This may work for 90% of the needs, but may fail to work with a life-video stream via Ajax events, thats why I try to use TWebBrowser in the first place., since this will work in the native environment with full performance. But thats another story. I will check such JS "injections" on Android first, then iOS later, but indeed I had hope to find a general solution to all platforms. This seems to be a lot of handcafted work, each on Delphi as well as on HTML side. Maybe there is the simple possibility to send events somehow, like TMessage to be received via TMessageManager) from JS to Delphi ?

Maybe Uwe's MMX-Explorer might be a helpful tool clearing your units up. (be aware, I'm not sure if its already 100% Rio-ready yet, better check before)

Dear Eli, thank you very much for sharing and finding such nice solution. I will check later, but that looks good, seems be able to inject something into the browser.

So you're looking for a pointer as alias 🙂 Maybe this will be the next language extension

Maybe be toggling the Visible flag for all entries from below the "Clicked" Header item to next Header item ? Certainly not nice, I know, but I think there is no grouping function included in FMX.ListView AFAIK. Not sure howto force Headers to fire events.

Do you mean like that ? https://community.embarcadero.com/blogs/entry/adding-headers-to-tlistview-programmatically This is done via TListItemPurpose.

Can you resolve this by building an empty test-app with XCode ? Usually XCode fixes (or at least warns) all issues, like missing/wrong provisionging files. After empty XCode app is running, you could try another empty app via Fmx.