A few of us have been getting together every Saturday and Sunday for the last ten or so weeks and playing around with live coding in Delphi.
We did this mainly to show some live coding in Delphi of various techniques as well as how to do things like a REST server/client at scale but also to show off some things like TMS Web Core which is a relatively new way of producing web apps using Delphi as the coding language.
This has taken place on Craig Chapman’s ChapmanWorld youtube channel. If you subscribe and hit the bell icon you’ll be notified the next time one of us streams. If you’ve not seen any of Craig’s videos/podcasts before you should pop on over. Craig and I are friends going all the way back to the late 90’s early 2000s when we used to both live in the UK and would rock up at the excellent Delphi User Group meetings at the POSK center in Hammersmith. Times have changed a lot since then. Craig and I are both now living in the USA and married to American wives with American kids. I swear this is something neither he nor I would believe if we could time-travel back to those days in London and tell our 90s ourselves that this would happen!
Others on the videos include Frank Lauter from Germany, Andrea Magni from Italy and I think even Jim McKeeth from Embarcadero in the US also popped in at least once. I apologize if I’ve not name-dropped others (for example, Gus and “The other Ian” both turned up to play almost every week) – in total we’ve collectively streamed 14 FOUR HOUR long episodes of live coding in Delphi.
We’ve been coding a card game and it’s taken ages, but we’ve kind of got to the point where we’re all going to move on to streaming about some other subjects. I will stream and finish the TMS Web Core client part of the game since we’re nearly done with it and it wouldn’t be right to just walk away.
A couple of notes about what’s going on in the stream. I am using TMS Web Core 1.3 running in Delphi 10.3.3 for the code. There is actually a version 1.4 of TMS Web Core out but I didn’t want to swap to using that as it was only released in the previous week and it does do a few things differently – although it also uses a later version of the Pas2JS cross-compiler which would have been helpful. Also, if I had been writing the server side (Frank and Craig did most, if not all of that) I would have used TMS XData and then linked the web client to the data using that in a cross-platform kind of way. I might do that on a stream soon just to show how quick that would have been. 20/20 hindsight is a wonderful thing 😎
Oh and sorry about the glitches with the audio in the first 20 minutes or so. Craig – who directs and hosts the stream – unwisely got some swanky new audio interface stuff and then found out he needed a PhD in audio engineering to actually use it in anger. Stick with it, it gets sorted out as time goes on.
Anyway, I hope you enjoy watching a bunch of Delphi appear before your very eyes on a live stream. We want to do more of this – Delphi is just TOO good at being the invisible driving force behind a lot of products and applications out in the real world and a bit of live streaming showing what Delphi can or can’t do I think can only be a good thing.
So many good things coming in the next release of RAD Studio, especially Delphi 10.4. A ton of stuff will crop up in the next few weeks; blog posts, demos, videos showing off delicious things like managed records, the TTitleBarPanel component and many other goodies.
You’re going to hear a lot of good things about Delphi 10.4 – after all, it is packed with lots of new goodness; fixes, new components, tweaks, managed records (you’ll learn about these and when you understand, you’ll totally get why they rock) – but above all else the Shout From The Rooftops new feature is the implementation of a Language Server Protocol or LSP.
Why do you care about this?
Well, it’s not just that it massively improves the code completion and other in-editor helper functions which in turn improves YOUR productivity when writing code..
No that’s not it;
Nor is it that the “red squiggles” (error insight) will now vanish in a puff of smoke when there’s no error – or that they will remain and be a truthful indication of the ACTUAL state of your code’s correctness when there is a genuine problem
No, not that either.
It’s THE SPEED AND RESPONSIVENESS.
The compiler is already stupid, crazy, fast (you get used to this and it’s a shock if you ever dip your toes in another compiler’s sludgy response and teabreak compilation speed) – but LSP takes all this to another level. FAST FAST FAST.
The sad part is that LSP is so good at just getting on with the job at hand people might not notice in a few weeks time that this is the Oeuvre triumph of the 10.4 compiler team.
If you have access to the RAD Studio Welcome page you’ll see me on it this week taking part in the EMB “Fireside Chat” series with the inimitable Jim McKeeth.
If you don’t have access to RAD Studio or the Delphi IDE then you can view the video directly here:
We cover quite a broad number of things such as fashions in programming (yes they exist, honestly) and converting legacy projects to the latest version of Delphi along with the challenges that brings plus the pros and cons of FireDAC and AnyDac amongst others.
Are you at a loose end and like watching people code live and enjoy hearing coders chatting about code and generally hanging out? Then you might want to subscribe to the Apocalypse Coders live stream.
It’s streamed live on Saturdays and Sundays – sometimes for as long as four hours – and features a group of Embarcadero MVPs and Delphi coders creating a cloud-based card game live on the stream using a multitude of technologies.
I’m one of the MVPs on the stream and you’ll see us using all sorts of coding methods and techniques to produce the server-side services as well as several versions of the client-side apps to showcase various technologies including VCL, FMX/Firemonkey and TMS Web Core. We intend to produce clients for Windows, Mac and Linux as well as a browser-based web client and we might even stretch out and produce iOS and Android clients too if the mood takes us.
We’re not intending to get too serious, it’s more to stream coding in Delphi rather than a huge formal tutorial, but even so we might lay out a few techniques which fellow coders may not have seen or wanted to know more about.
One of the MVPs, Andrea has produced a playlist of the most recent stream as well as an online Delphi Italia event here:
I’ll be there again this Sunday at 9am CST, 8am MNT, 14:00 UTC / BST – you can subscribe to Craig’s ChapmanWorld YouTube channel and get notified when we go live and catch the stream by going here: https://www.youtube.com/c/ChapmanWorldOnTube
If you’ve been using Delphi for a while you’ve almost certainly come across the fabulous TMS Software with their enormously useful and comprehensive set of components. If you haven’t heard of them yet it’s worth clicking on the link and taking a look at what they have to offer – you’ll definitely end up saving yourself a massive amount of time using something ready-made from TMS rather than writing a whole bunch of code of your own.
Anyway, recently, Craig Chapman, Frank Lauter and myself have recently been streaming live coding on Craig’s Chapmanworld YouTube channel. For the last couple of weeks we’ve been writing a game loosely based on Apples to Apples and/or Cards Against Humanity type games.
This week I took over quite a large part of the streaming time to show TMS’ Web Core components which allow Delphi programmers to produce fully-capable web apps from within the Delphi IDE which are then transpiled into pure HTML and CSS. That’s right: write Delphi object Pascal code, design the forms and HTML pages in the IDE in the normal way – and out pops a real self-contained web app all ready to go.
The stream is quite long, 4 hours, but I start getting down to actually using TMS Web Core and writing code around 1 hour 6 minutes in. The embedded video below starts from that point onwards although you can wind back to the start of the stream if you want to hear Craig, Frank and myself discussing various things (and getting the kinks out of the streaming technology).
We’ll be streaming every Sunday at the same time at least for the next few weeks – 8am CST, 2PM BST, 1pm GMT. Go to the Chapman World web pages for more details and subscribe to the YouTube channel to get notifications of when we’re next live again.
Incredibly it has been 25 years this Valentine’s Day since Delphi was first released. I read a pre-release magazine article about it and it was completely obvious that this was going to be a whole new way of doing things for me – and all for the better too.
Now, 25 years on, I am an Embarcadero MVP for the USA and I spend every single day coding in Delphi. It has for the last couple of decades been the primary source of my income. But it doesn’t just stop with me – it’s also chugging away behind the scenes working to provide services for millions of employees in the UK through my apps and I have software in 66 different countries translated into almost as many languages.
It’s a pretty wonderful thing. If I could travel back in time to when I was 14 years old I could show Younger Me all I have achieved as a software developer, the countries I’ve visited and even amaze my teenage self by telling him how I no longer live in rainy England but instead bask in the Texas sunshine in Dallas – and as a US Citizen too with an American wife and daughter. Mind blowing. Delphi has played a consistent and on-going part in how my life has turned out, for the better.
As part of this week’s celebrations a few of the MVPs, myself included, have made videos talking about how they were influenced by Delphi and what it means to them. My video, which includes some personal history and background to the work I do, appears above.
As in all things there are fashions in UI and UX design. Right now perhaps the biggest trend is “dark mode”. Unless you’ve been living under a rock you probably know this is where the operating system – Windows, iOS, macOS and Android – supports a user interface theme where the elements can be set to a “dark mode” which generally means all those big swathes of white automagically turn to a dark charcoal or black color.
Windows 10 supports such a mode. If you’re running one of the later builds you’ll find that all sorts of things can be ‘dark’.
I personally run all my devices – I use Windows, macOS, iOS and Android every day – in dark mode. I suffer from a harmless but annoying visual issue which people commonly call “floaters”. It’s a factor of my age and eyesight. This means that bright white screens are not optimal since these little whispy things float around in my field of view while I am trying to get on with the job of coding. With Dark Mode enabled the floaters, which are also dark, are almost completely mitigated to the point where I don’t even notice them.
So, for me, and perhaps others, Dark Mode is a huge benefit rather than just an aesthetic choice and I want every app I can to support it.
This is where my CodeRage 2019 video (below) comes in. We can detect whether Windows 10 is running in Dark or regular ‘light’ mode using standard and fairly straight-forward Delphi code. When we know what mode the OS is running in we can use the power of Delphi’s VCL to swap the entire visual experience of our application to an appropriate theme.
In the video I show you how to do this with a simple Delphi unit I wrote which you can include in your own apps. There’s also a demo project you can play with.
One thing I don’t discuss is what to do if your app is already running and then the user changes the Windows theme from light to dark or vice versa.
To cope with that scenario you need to trap the WM_SYSCOLORCHANGE Windows message.
When the WM_SYSCOLORCHANGE message is triggered you then need to check the Windows theme mode and react accordingly. Like so:
Note that all of this pertains to Microsoft Windows – and Windows 10 onwards. Other OS do it differently. Note also that Delphi 10.3.3 Rio has just been released and this includes specific code for Dark mode in your apps.
before we start, note that this article describes code-signing on Microsoft Windows for applications designed to run on Microsoft Windows 32 or 64bit. Apple’s macOS, iOS and the Android OS also use code-signing but this process is different and uses a different chain of incompatible tools.
Open the pod-bay doors
We all know that writing code is a non-stop smorgasbord of fun, adventure games and caffeine-induced migraines. But eventually the party music has to stop and we need to get our programs out into the sweaty hands of the unsuspecting general public.
Or as Embarcadero (and Microsoft) call it: deploy.
Deployment of applications back in the halcyon days of Windows XP really meant “copy it to a CD” or, if you’re a “mature” programmer like myself maybe even a floppy disk. Oh such happy, carefree, virus-laden days.
Now of course operating systems have become a lot more sophisticated and gained complexity in deployment thanks to the preponderance of computer viruses, worms and trojan apps.
The old ways of trusting our luck just don’t work any more and the virus-writers are almost constantly engaged in a digital arms race against the anti-virus and operating system vendors.
Add to that the huge improvement in connectivity with fiber-optic cables or super-fast internet being ubiquitous in most of the Western World and gradually to the neglected rest too, then installation of applications is often digital – a link on the internet – rather than via some physical medium.
It’s a matter of trust
With easy digital distribution problems started to crop up. The hackers and crackers started to spoof URLs of popular application vendors. More than once popular applications were copied and injected with viruses. The spoofing got so good that it soon became very difficult to tell if the program you had downloaded was actually from the genuine vendor or from some nefarious bandit chomping Cheetos in a sweat-laden bedroom viper pit while he/she scooped up your banking details.
Enter a solution – code signing. Code signing uses a mechanism to embed a cryptographic-protected digital ‘certificate’ into your application’s exe or package. The certificate is issued by a Certificate Authority which is one of a handful of well-known, publicly-listed companies. The certificate is issued to the application vendor using a secure mechanism after which the application vendor (that’s you!) then uses one of a selection of special tools to sign their applications with the certificate.
This ‘signing’ embeds the certificate into the application’s exe or executable module. Once embedded in this way the certificate or any other part of the signed application file cannot be changed without breaking the certificate. This is important as it stops viruses from infecting your application and then giving the user a false sense of security because they trust you and the appearance of the certificate means you and only you created that application module. Note that it doesn’t mean you can’t accidentally code sign an application module which has already been infected with a virus or contains trojan code – but it does mean everyone will know you did it. Of course, I’m sure you use several layers of anti-virus protection on the computers which you use to build and deploy apps… right?
With the advent of Windows Vista the Windows operating system introduced User Account Control started to look for these embedded certificates and if the application you were trying to launch didn’t have a certificate Windows would show a slightly fear-inducing message telling you the publisher was unknown with a passive-aggressive hint that this would be a Bad Idea
Once you code-sign your app the sick-yellow warning gets replaced with a much more comforting blue-skies thumbs up:
As you can see in the above dialog – my name appears by “verified publisher”. This is because this was the name of the publisher on the code signing certificate. So it’s important to pick the right name if you go to purchase one.
Now, signed like this no-one can be in any doubt over who created this application because it was me who signed it or at least someone I gave the code signing certificate to along with the credentials to use it.
How to get a code-signing certificate
There are a few types of code signing certificate. Note that none of them are the same as SSL certificates used for web sites.
This is a confusion which seems to crop up regularly – the two things are similar in nature but used for different purposes and you can’t use an SSL certificate to sign code and you can’t use a code signing certificate to enable SSL and make the padlock appear on your website.
The main types of code signing certificate are:
You create this yourself and there is even a button in Delphi 10.3 Rio’s provisioning page when you select a distribution type of “Windows Application Store”. It has the least amount of ‘trust’ and is used either for internally-distributed apps (for example within your company) or for testing purposes. You can’t use this type of certificate for ‘regular’ apps that you want to distribute to large numbers of people and even among friends you may find they are unable to run your app as they might have Windows settings which prevent this; in particular Windows 10 S is a particular flavor of Windows, installed on Microsoft Surface devices by default, which prevents installing any apps at all unless they arrive via the Microsoft App Store. In short, self-signed certificates are only just better than unsigned apps. They do have the advantage that they’re easy to create and they’re free but that’s about it.
OV Certificates – or “organization validated” certificates are the most common and cheapest form of code signing certificate. This certificate costs money. The OV certificate requires the person in whose name the certificate will be issued to go through a verification process. This process can take a day or two – sometimes longer depending on what information the certificate issuing authority asks for and how easy it is for you to obtain.
For registered companies such as Ltd, PLC, LLC, Inc and charities the verification process usually uses publicly-available information such as Dun & Bradstreet and other public legal listings to validate the existence of the organization. There is usually some extra machinations such as providing a letterhead and taking part in an automated verification phone call to a number listed in the public records for that organization. It’s usually straight-forward.
For individual developers it’s a more involved process. In my case when I wanted a code signing certificate in my own name – rather than a company – I had to take various proofs of identity to a Notary Public. This is because I am based in Dallas, Texas, USA. Developers in Europe and other countries will have a different but similar process. The notarized copies of ID (there was more than one document) were then sent by the notary to a specific fax number and scans were emailed too. The whole process took about a week. I used my local UPS store for the notary services – they were cost-effective, less than $10,and I use them now for all notarization (which in the USA seems to crop up a lot!)
Despite what you may hear in most cases an OV code signing certificate is perfectly fine for the job. I write code for a number of software houses and one in particular signs all app resources using an OV certificate. These apps go out to in excess of 10,000 active customers in a huge variety of operating system configurations and deployment scenarios from regular Joe public to small offices to blue chip public listed companies including several airports and at least one Embassy that I am aware of, maybe more.
To code sign with an OV certificate you can use a utility I will describe later or Microsoft’s own MSSign application. Delphi 10.3 Rio also can sign your apps too with the OV certificate as long as you’re targeting Ad-Hoc on Windows.
For a start the validation process is much more rigorous. Once you pass this validation process you will be issued with a USB hardware ‘key’ which is a device that acts as a ‘token’. This token must be physically present each and every time you wish to sign some code. I know that for some people this has proved to be a problem and can make automated builds a pain since you have to physically type in a password to authorize access to the token – a step which apparently can’t be fully automated.
Once you do obtain an EV certificate it does mean that things like Microsoft Smart Screen will provide you with the highest level of trust. For some scenarios this is going to be an absolute requirement.
For me, I’ve not found it necessary to get anything other than an OV code signing certificate – I recommend you read through the links above and make your own choices.
Microsoft App Store Code Signing
This final type of certificate is a bit of a weird one. When you submit an app to the Microsoft App store – either paid or free – Microsoft pass it through a validation process which checks that you’ve not used any forbidden API calls or included any other obviously infringing content. Once you pass that validation Microsoft will then sign the app for you. The code signing still identifies the author as you and is based on the “publisher” details you provided to Microsoft when you signed up to their App Store as a developer and submitted the package. I’ve yet to work out quite how Microsoft validate this as strongly as an EV or OV certificate but in the end I guess it’s not relevant to you as the developer.
Where to get a code-signing certificate
A disclaimer – I don’t have any association with any of the companies I’ve linked to here and elsewhere in this article. I can vouch for K Software because I’ve used them for a few years now for several customers of mine and always get my certificates from them but beyond that I don’t know them and they don’t know me 🙂
The cheapest/most cost effective certificate vendor I’ve found is https://ksoftware.net – they are resellers for Sectigo (previously known as Commodo) and provide both OV and EV certificates. The renewals are discounted too. I assume they survive off Ramen noodles for lunch every day. Either way, I’ve used them for a while and they’re still going strong so more power to them.
The number one main source for code signing certificates is Sectigo (Commodo) themselves and can be found here: https://sectigo.com/signing-certificates/code-signing – weirdly enough they are a lot more expensive than K Software who resell their certificates. I’m not sure why this is.
There are a large number of other resellers and top-level certificate authorities. I do not have experience of them so I’ll leave you to wade through the morass. Don’t pay too much, is my advice, but make sure you’re getting the correct thing.
How to code-sign your apps
I’m not going to describe the process of actually getting the certificate because the vendors do a good job of walking you through it. But however you do it you will end up with a file which either ends in .p12 or .pfx – here’s a pro tip: some apps and techniques for code-signing ask for P12 files and some want a PFX file. Whatever one you have make a copy of it and rename the copy’s extension to whatever the other one is. So P12 to PFX or PFX to P12. The reason this works is they are exactly the same file format at the binary level. Only the file extension is different. I know of one person who was going to pay for a second code signing certificate “because they had the wrong format”. Luckily they ranted in my direction and I was able to save them doling out cash unnecessarily.
If you have an EXE and you simply want to code sign outside of the IDE
No matter where you got your certificate, go to K Software’s website and scroll to the link which says Download KSign Now. KSign is a small Windows app which allows you to manually code sign any of your exes.
The app has areas for the certificate details and an area where you can drop or select one or more exes to code-sign. It’s easy to use.
Manually signing using Microsoft’s SignTool
You can use Microsoft’s command line code signing tool. This may be already installed on your machine.
It is very common now that users have more than one monitor; even laptop users often dock a screen which gives them two screens. People even sometimes use tablets as second screens. All of these can use different resolutions.
Microsoft caught on to this trend and with successive versions of Windows 10 they introduced the “Per monitor v2” technology. This allowed users to have different DPI scaling for each monitor. Now, for example, users could have a laptop’s LCD panel showing at its native resolution and an additional external monitor displaying at perhaps a higher resolution of DPI. Note that this is not just the pixel width and height but the actual denseness of the pixels per inch.
If you dragged your legacy Delphi apps from one window to another you could see that change in pixel density. Things would look sharper or less sharp, scaled weird or, even worse, drew text in the wrong places because the coordinates were worked out on a machine running on and launching at 96DPI and was dragged over to a screen now running at 120DPI.
Prior to Delphi 10.3 this made our apps look bad. Not so now.
Rio to the rescue…
In Delphi 10.3 Rio there is now some settings found in the main menu by selecting project, options and then clicking on manifest. You can also press SHIFT + CTRL + F11 to get to the same place.
The red lines in the image above show you where to check. Under DPI awareness you should have Per monitor v2 selected. For new projects and those upgraded from legacy versions of Delphi you should find it’s already been chosen for you.
In addition to the Delphi runtime taking care of enabling Per Monitor v2 by notifying Windows that our app understands it there is also a new event which gets triggered on your forms if the screen’s DPI changes, perhaps due to them changing the Desktop resolution in the Windows preferences.
This event is OnAfterMonitorDpiChanged and it triggers just as your would expect.
Other options on the project options page
While we’re on the project options page, just above the setting for Per Monitor v2 you can see an option which says “Enable Runtime Themes“.
This setting needs to be checked for the themes to work (it’s checked by default). We’ll talk more about runtime Themes in a later blog post. This setting actually does a similar thing to the old TXPManifest component which has been around since the days of Delphi 7. It essentially includes extra information in the application’s manifest which tells it to render native controls in a particular way. I’d recommend you have it checked.
Execution level is another setting on the same page and is actually a pretty important feature too. It controls the user permission level at which the app runs – default is “as invoker” but you might not want that. This setting tells the Windows runtime what execution privileges you wish to run with. If you select “as invoker” then you inherit the rights of the shell or app which launched your app. So if the shell launching your app is very restricted your app will be restricted too. Conversely if the launching process is, for example, ‘administrator’ then you’ll inherit the rights which come with that. Sweet. The two other options are “Highest available” and “Administrator” which are fairly obvious – but both of these might need a normal mortal user to enter an administrator password or credentials before your app can run. In most cases you’re going to leave the setting at “as invoker”.
Next article – provisioning, Windows application store and code signing
In the next article I’m going to cover the new Delphi 10.3 Rio features which allow you to package your Windows apps for the Microsoft App store as well as a discussion about code-signing; what it is, how to do it and where to obtain a code-signing certificate.
The great thing about Delphi is that it’s totally rock solid. I mean bullet proof, runs for ever, doesn’t need a bunch of dependencies, keeps on going for ever solid.
Because of that there’s a lot of legacy code out there. Many millions of lines of Delphi code, quietly doing its thing without fuss or drama. That’s great, but of course, the world keeps on spinning and eventually Moore’s Law means that while our programs work the environment in which they operate has changed.
Often these changes are in fashions of user interface look and feel such as new colors and the adoption or abandonment of passing trends like skeuomorphism or super-flat designs mean our older apps start to look very dated, often breathtakingly quickly as a new version of Windows wobbles its way down the now-ubiquitous always-on internet.
One of the most subtle changes to come about is the advent of 64bit Windows. For many people it happened silently. They bought a new laptop (75% of computer sales in 2018 were either laptops or tablet devices) and that laptop, in the last few years at least, ran a 64bit version of Windows 10.
Now, increasingly, I am finding that many of our corporate customers are looking to only run 64bit versions of apps. This means, if your app or particularly Windows service, is not 64bit then the customer will not allow it on their system. I did a rough count earlier via our support techs training or installing today and it was 3 of 10 servers who expressed either a preference or mandated that the system be 64bit only.
So, your Delphi 7, 2009, XE 1, 2, 3 etc apps are in trouble as they are using 32bit architecture.
Luckily, Delphi 10.3 Rio makes this an easy thing to overcome. Right click on the project in the project manager window and select “add platform”. Now choose “Windows 64bit”.
If you’re running a 64bit version of Windows you can go right ahead and select “build” and, if your components are 64bit compatible, you’ll end up with a pure 64bit exe. Simple as that.
If you’re using a 32 bit version of Windows you’ll need to connect to PAServer on a 64bit machine. To check your Windows version press the Windows key and ‘X’ and then select “system”. It will say “32 bit” or “64bit” like it does in the image at the top of this article.
But I should add a caveat; not all older components can work in a 64bit architecture. Encryption, for example, often is tied quite closely to the processor ‘bitness’. You may find that older code of this type either needs to be tweaked or replaced to work for 64bit Windows.
Likewise, some component vendors annotate their components to target only 32bit Windows, either due to the use of APIs which are incompatible with 64bit or due to other technical problems such as the size of integers and so on.
You can check if a particular component will work when targeting Windows 64bit by pointing your mouse at the component in the component palette in the RAD Studio / Delphi IDE. A hint window will appear listing all the target types on which the component can be used: Windows 32bit, Windows 64bit, iOS, Android, Linux, macOS and so on.
Here’s what the hint looks like (in dark mode) for a component which is only available for Windows 32 and 64bit.
We’re going to look at another silent feature of Windows 10 which will make your legacy programs look long in the tooth: high res DPI and the spectacularly-named “per monitor v2“. Come back tomorrow to see what it’s all about.