Lars Fosdal

To write an expert that helps separate the UI from the logic would be far from trivial. I strive to write code that is testable, something that Steve McConnell's "Code Complete" drilled into me. My logic routines and biz objects are created without a UI, and the UI itself is connected afterwards. I make an effort to isolate the displayed data from the data that is actually used. I.e. a grid only reflects the underlying data, and all work is done on the underlying data and not the data in the grid itself. A checkbox state reflects the underlying state, and a change to the checkbox affects the underlying data - which then again applies any recalculations to states, and suggest a refresh of the display. The benefits: It is easy to write unit tests and integration tests for the biz logic. It is easy to reuse the code, or change the UI. It also simplfies changing the biz.logic as it is inside the "black box" object, and not heavily tied to the UI. If the tests says the object works, and the app doesn't work - the error is in the coupling between the non-visual object(s) and the UI. The drawback: There is a certain degree of data duplication - but the cost of that is quite limited. You can use proxy classes and/or Attributes/RTTI to associate elements between the non-visual and UI elements. Divide and conquer is good - but hard unless applied from the start on.

http://docwiki.embarcadero.com/RADStudio/Rio/en/TeeChart

Have you setup the required permissions in the Android manifest? From https://developer.android.com/reference/android/telephony/PhoneStateListener#onCallStateChanged String: call phone number. If application does not have READ_CALL_LOG permission or carrier privileges (see TelephonyManager#hasCarrierPrivileges), an empty string will be passed as an argument.

https://docs.microsoft.com/en-us/windows/win32/perfctrs/using-the-pdh-functions-to-consume-counter-data

Wow! That fact had eluded me!

The book gives a reasonably good understanding of threading issues. That understanding would suggest alternate solutions that avoids the pitfalls of messaging.

A good place to start for writing solid thread code: https://leanpub.com/omnithreadlibrary

Ref installation: You can safely install without elevating to localadmin.

If you had TMS libs from before, like the TMS Component Pack, those are purely VCL for Windows. There are currently two alternatives for cross platform: 1. FireMonkey + TMS FMX Components https://www.tmssoftware.com/site/products.asp?t=fm (All Firemonkey supported platforms) 2. CrossVCL https://www.crossvcl.com/ - not compatible with TMS for VCL, afaik. (MacOS and Linux) Disclaimer: I've not used any of these, only the TMS Component Pack for VCL.

Using the std TJson class in REST.Json. Probably not considered gold std.

Yes. I use the thread class name and an instance id.

Also, monitor your memory consumption in the Windows App. If it is not stable but keeps growing - there is a chance that you have a leak.

That is never wrong. Please ignore me 🖖

Not sure why one would want to do this in a Delphi app, when the tools are right there in PowerShell. https://docs.microsoft.com/en-us/powershell/module/microsoft.powershell.management/get-hotfix?view=powershell-7

Firefox? I assume Firemonkey. You are more likely to get help if you are specific with regards to what you want to do.

I know... My indendation style is unconventional. It works for me, though.

Scoured our complete source code and found only one occurence each of pred and succ. One was legit. for e := Low(TElementType) to Pred(High(TElementType)) do // i.e. do not include the last enum value in the loop. The other one was a bullshit one - and I was probably the author of both. function AfterFirst(const Match:char; const st:string):String; var index : Word; begin index := pos(Match,st); if index <> 0 then Result := Copy(st, index + 1, succ(Length(st) - index)) else Result := st; end; It's not even consequent.

How about using MadExcept or EurekaLog ?

That said, there is BigInteger from https://github.com/rvelthuis/DelphiBigNumbers

An IPv6 address is 128 bits. The answer is no. You'd need two UInt64s to hold that.

But isn't it still a value type? https://docs.microsoft.com/en-us/dotnet/csharp/language-reference/builtin-types/nullable-value-types#how-to-identify-a-nullable-value-type

100% guesswork without any foundation, except name likeness, follows: There is a GetUnderlyingType property in System.RTTI type TRttiEnumerationType - but it is used absolutely nowhere in the RTL. Also, WinAPI.Ole2 contains a VariantChangeType, and WinAPI.ActiveX contains a VariantChangeTypeEx that may be relevant. System.VarUtils contain both. function TRttiEnumerationType.GetUnderlyingType: TRttiType; begin Result := Pool.TypeOrNil(TypeData^.BaseType); end; function TRttiPool.TypeOrNil(Value: PPTypeInfo): TRttiType; begin if Value = nil then Exit(nil); Result := GetType(Value^); end; function TRttiPool.GetType(ATypeInfo: Pointer): TRttiType; begin if ATypeInfo = nil then Exit(nil); Result := TRttiType(ReadObjectPointer(TRttiType, nil, ATypeInfo)); end;

I see. Personally, I rely on the MRU, Quick Access and Libraries functions that are built into the File Explorer. They are sufficient for me.

I am not sure if these are supported in Lazarus, but you could try an ActionMainMenuBar + ActionManager where you only use the top-level menu items. It would handle the hotkeys for you. ActionBarDemo.zip

@limelect What is the nature of the interaction between your app and the file explorer? Would it be better to launch your own explorer with CreateProcessEx? That way, your app would be the owner, and you would know when it closes, and if you need to launch another. In Windows 10, you can control if folder explorer windows should have separate processes by default or not. The default appears to be not, meaning each explorer opened (from f.x. using Win+E) actually is running in the same process. What is your setting? Does changing this setting affect your observable problem?