Hot Reload in Delphi?

[PeterPanettone 157]

Hot Reload is a feature in the upcoming .NET 6 that allows developers to modify source code while an app is running and immediately see the results:

 

Microsoft reverses controversial .NET change after open-source community outcry

 

Just pipe-dreaming: Wouldn't that be a wonderful feature for Delphi?

[Lars Fosdal 1792]

It would be a nice feature, but it is pretty unlikely to ever happen.

The .NET platform is very, very different from Win32 on Intel.

 

The ancient Borland Pascal DOS debugger had backstep, where you could single-step backwards from a breakpoint.

Also not quite as easy in the multithreaded, asynchronous world of Windows.

[David Heffernan 2345]

14 hours ago, PeterPanettone said:

Hot Reload is a feature in the upcoming .NET 6 that allows developers to modify source code while an app is running and immediately see the results:

 

Microsoft reverses controversial .NET change after open-source community outcry

 

Just pipe-dreaming: Wouldn't that be a wonderful feature for Delphi?

Great to see that MS has done the right thing here, although disappointing that it took so much pressure to make it happen. Good on El Reg for their reporting on this. 

 

Never ever going to happen for Delphi though!! 

[Uwe Raabe 2057]

18 minutes ago, Lars Fosdal said:

The ancient Borland Pascal DOS debugger had backstep, where you could single-step backwards from a breakpoint.

The current debugger lets you simply drag the execution pointer to a line before or after the current position:

[Fr0sT.Brutal 900]

1 minute ago, Uwe Raabe said:

The current debugger lets you simply drag the execution pointer to a line before or after the current position:

Nice! I was always using RMB>Debug>Set next statement

[Stefan Glienke 2002]

They cannot even build a stable debugger - how could they ever achieve this.

[Anders Melander 1782]

26 minutes ago, Uwe Raabe said:

The current debugger lets you simply drag the execution pointer to a line before or after the current position

I don't think I've ever had a situation where that didn't just mess everything up.

I think it just adjusts the EP so the other registers will not contain the values that corresponds to the new EP.

[Anders Melander 1782]

34 minutes ago, David Heffernan said:

Great to see that MS has done the right thing here

Why was their original decision "the wrong thing"? Wasn't it simply a business decision that a (maybe large) group of people was dissatisfied with?

[Lars Fosdal 1792]

32 minutes ago, Uwe Raabe said:

The current debugger lets you simply drag the execution pointer to a line before or after the current position:

Most of my code involves a multitude of objects in various containers and their interactions.

I've never dragged the exe pointer or set the execution point during debugging, as it would most likely create conflicting data.

[Lars Fosdal 1792]

1 minute ago, Anders Melander said:

Why was their original decision "the wrong thing"? Wasn't it simply a business decision that a (maybe large) group of people was dissatisfied with?

The "wrong thing" was to first introduce it to the open source libs, then pull it back - instead of keeping it private, and adding it to the open source lib later.

[Anders Melander 1782]

17 minutes ago, Stefan Glienke said:

They cannot even build a stable debugger - how could they ever achieve this.

I know you're just venting, but I can't see how they could achieve it even if the debugger was better... or why they should.

[Anders Melander 1782]

3 minutes ago, Lars Fosdal said:

The "wrong thing" was to first introduce it to the open source libs, then pull it back - instead of keeping it private, and adding it to the open source lib later.

Makes sense. Thanks.

[David Heffernan 2345]

23 minutes ago, Anders Melander said:

Why was their original decision "the wrong thing"? Wasn't it simply a business decision that a (maybe large) group of people was dissatisfied with?

I think this controversy was well covered at The Register, I'd read their articles if you are interested 

[Lars Fosdal 1792]

The Register is definitively one of the more interesting and accurate sites on all things IT these days.

Very little fluff and rumors.

 

Edit:
Here is the "wrong thing" article: 

https://www.theregister.com/2021/10/22/microsoft_net_hot_reload_visual_studio/

 

Also, the BOFH posts usually are good for a laugh and a nod.

 

[David Heffernan 2345]

1 minute ago, Lars Fosdal said:

The Register is definitively one of the more interesting and accurate sites on all things IT these days.

Also, Verity Stob!

[Lars Fosdal 1792]

6 minutes ago, David Heffernan said:

Also, Verity Stob!

It has been too long since the last "Clan of Kahn" "Sons of Kahn" post 🙂

[David Heffernan 2345]

Just now, Lars Fosdal said:

It has been too long since the last "Clan of Kahn" post 🙂

Perhaps this is just the final confirmation of the end of days for Delphi, if Verity Stob isn't producing new Sons of Kahn material, then does Delphi really exist?

[Fr0sT.Brutal 900]

1 hour ago, Anders Melander said:

I don't think I've ever had a situation where that didn't just mess everything up.

I think it just adjusts the EP so the other registers will not contain the values that corresponds to the new EP.

I must say a word of defense here. Probably my cases were simpler but I successfully do this "step back" from time to time.

[Mike Torrettinni 198]

16 hours ago, PeterPanettone said:

Just pipe-dreaming: Wouldn't that be a wonderful feature for Delphi?

At first I couldn't imagine why could this be useful. But then I thought about all the little visual changes I need to do to on some forms to fit the desired look, and being able to hot reload a simple change of position or alignment or text and no need for reloading the project, this wold be really cool to have.

Not sure how could it be useful for anything more than simple changes, but even just for my use case, I would vote for the 11.1/11.2 patch to include this feature!

[Stefan Glienke 2002]

1 hour ago, Anders Melander said:

I can't see [...] why they should.

Because hot reload is a modern part of the R in RAD. Stopping, recompiling and getting to the point where you were before the change would be completely unnecessary. Game devs using UE or similar tools have been using this for quite a while.

[Martin Sedgewick 30]

I remember years ago watching a video of "Notch" using Java to create a game and he leaned heavily on hot reload.

 

It was immediately obvious the benefits. The game was running. He modified some function or other and hit a key and the game updated with the change. The time saved over days, weeks and months would be incredible. Obviously it depends on the software. but if you have to compile, run, login, get to point X it all adds up.

 

[Anders Melander 1782]

21 minutes ago, Stefan Glienke said:

Because hot reload is a modern part of the R in RAD. Stopping, recompiling and getting to the point where you were before the change would be completely unnecessary. Game devs using UE or similar tools have been using this for quite a while.

I'm not questioning the usefulness of it. I would absolutely love that feature. I'm questioning why Embarcadero would want to invite the inevitable complaints when the feature, inevitably, failed to work 100% of the time - or even just not as good as VC/VC++.

 

AFAIK, what UE does is just reloading dynamic DLLs if you recompile them. Not really comparable I think.

[Uwe Raabe 2057]

55 minutes ago, Anders Melander said:

what UE does is just reloading dynamic DLLs if you recompile them.

That is basically what you already can achieve with Delphi when you work with dynamic package loading. The point with such an approach is that the code must be able to store and restore states between a reload.

[dummzeuch 1505]

4 hours ago, Anders Melander said:

I don't think I've ever had a situation where that didn't just mess everything up.

I think it just adjusts the EP so the other registers will not contain the values that corresponds to the new EP.

I use it often to repeat a call to a function where I missed something when stepping trough the first time. Works without a problem for this and saves the time to restart the processing. e.g.:

SomeCodeHere;

try
  SomeFunctionCall(SomeParameters); // <== Code raised an exception I need to investigate
except
  on e: Exception do
    LogException(e);
end;

SomeMoreCodeHere; // <== so I place a breakpoint here
                  //     (after the debugger stopped on the exception)
                  //     and then set the execution point back to the function call

Of course after that the call stack may be broken, but usually I stop the process anyway to fix the problem.

[PeterPanettone 157]

A great way to change things at run-time in Delphi applications is a RUN-TIME OBJECT-INSPECTOR.