Fr0sT.Brutal

Oh, you're right, thanks! Then helpers seem the most convenient way to go.

AFAIK there could be only one class helper until some of 10s so adding this you're blocking a user from creating his own ones. In this very case, you can define a descendant class where Count property would be declared for older compilers (GetSize is protected so there's no problem). Then you can typecast or just use type name as a pseudonym for your derived class: {$IFDEF Old_Compiler} TCompatJSONObject = class(Json.TJSONObject); property Count: Integer read GetSize; end; TJSONObject = TCompatJSONObject; {$ENDIF} If I get it right, non-published properties are "virtual" (just a syntax sugar) so these classes will even be binary-identical. In general, there's no convenient universal solution, every case should be considered individually.

... and Addr() 🙂

Yeah, they got it right. Though they have to keep Delphi compat. Could both interface types be used together? Very nice workaround!

Delphi interfaces mix two concepts - ref-counting with auto disposal and the very interfaces as a required method set - into one type whereas frequently only one of them is needed. While ARC was bad idea because it broke too much old code, it had some sense. I guess if Delphi introduced new types like IPlainInterface and TRefCountedObject, it would be pretty nice and useful

+1 for naming, it's handy. I do it manually though with just a class name

Anyway `While MediaInfoThread.IsSuspended = False do Sleep(1); ` looks very suspicious to me, like useless wasting CPU cycles.

I was bitten several times by errors in "for 0..N-1" loops when container was empty and iterator was unsigned. So for arrays "for Low..High", for other types "for in" and "for 0..Count-1" only when modification of an element is required

OK, now that I see the interface my suggestion of blocking seems ineffective (though it would be nice in another cases). In this case: - Launch render in bg thread generating some unique value that will identify its results (ExpectResID := NewGUID; Thread.ResID := ExpectResID; Thread.Start) - Output some information like "Render in process..." on the panel that will later contain the image - If user starts another render, launch one more thread with another ResID and override ExpectResID (the value of ID to expect) so that when 1st thread finishes and reports results, they will be discarded. Optionally, if rendering algo allows cancelling, cancel the old thread to release resources. Otherwise just let the thread finish its job. - When rendering thread finishes, post this event to form async-ly with PostMessage - Paint the image - Optionally consider limitation of number of launched threads

That really does address the cause of the problem. There's no sense in giving an opportunity to launch several renders at the same time. And blocking of operation's repeat is very common in UI

That's why I said "could". 😉

The coolest thing about for-in is that you can implement it for any custom class or record by implementing an iterator. So, f.ex., you can have `for jsonval in JSONDoc.Values['somearray']`. For-in works for strings as well iterating by chars though I personally haven't used it in such way. And it's a bit slower so for time-critical routines usual `for` still remains actual (though for very super time-critical things pointer increasing could beat `for` anyway)

Just disable ability to launch another render until current one is done.

SQL.Text := 'SELECT * '+ ' FROM ... '+ ' WHERE ...'

I agree adding of such comments is not a formatter's task. It would be useful in more advanced form, like generating of documentation stub with all parameters and result.

I only use Pred rarely with enums. With integers it's worse than N-1. I guess having for-in construction there's not much need in last index

They say " Anyway, OTF and TTF are pretty much identical file structures. OTF simply has more structures for enhanced font features. " in that topic.

AV is when you access memory not belonging to your process. In this case you just access memory that is being freed but still belongs to the process. If you want array index control, enable range check

Maybe you could start from this https://forum.lazarus.freepascal.org/index.php?topic=33141.0

Sure, he has 192.168.* which is local network address.

I guess you didn't really meant process

Yep, I know 😉 For my projects this fact pushed me to change most of in-place string concatenations to formatting of constant literals which additionally is good preparation for easy localization in future.

Too specific IMHO. Couldn't Delphinus help you in your case?

Noticed this hidden catch too so turned all my routines to use CreateFmt. Btw, the same goes for string concatenation raise Exception.Create('Error: '+ErrorText);

Sorry, I have no Delphi nearby currently, try to read ICS samples, there are plenty.