Delphi Dying? Think again, Tiobe
At the beginning of last week, Tiobe once again threw a punch at Object Pascal. Playing the whole “Delphi is dying” tune, while focusing on outdated and quite frankly irrelevant episodes from the past. Hoping no doubt, to leave the reader with an impression that Delphi is stuck in the 90s.
This is the same pattern we often see whenever Delphi or Object Pascal in general experience significant growth; or to be blunt, when the author cannot be bothered to think independently, but simply parrot hearsay and misinformation on autopilot.
It is lame, superficial and Tiobe’s biggest mistake to date.
Just to underline the problem areas here. The ranking is based on their internal system (there is no standard for how to rank popularity), and while I have issues with how they build up their score, it’s ultimately the March editorial text that has caused irritation and shock. You don’t declare a language as dead when there are over 10 million developers using it. This type of editorial could have very real consequences – which in turn brings us to their ranking system and how they arrived at their conclusions.
I would have understood their statement if it was issued between 2007 and 2010, because Delphi was at that time transitioning between Borland and Embarcadero. But to issue something like this in 2020? After a decade worth of restoration, optimization, modernization and above all – forging a thriving community that goes from victory to victory month after month, year after year? It makes absolutely no sense.
Significant growth
In 2018 there were roughly six million Delphi developers (I worked at Embarcadero at the time), with a total estimate of ten million Object Pascal developers worldwide when counting all alternative compilers, dialects and indeed – known piracy issues.
“Tiobe failed stupendously in their data mining operation, they seem to be oblivious regarding the demographic in which the language is used”
Since that time Delphi has made strides into the universities in Scandinavia, South-America and the Middle-East. Turkey recently announced their dedication to native and archetypal software development with Delphi (provided free for students), which adds a whopping one million students to the already large body of users.
Embarcadero has slowly but steadily rebuilt much of the infrastructure that existed under Borland. From professional training at Embarcadero Academy, to entry level training at LearnDelphi.org. The Idera community pages likewise produce a large body of articles on a weekly basis. Comparing the Delphi and C++Builder ecosystem today with it’s tragic state back in 2010, is like day and night.
With so much positive happening in the world of Object Pascal, Tiobe’s article comes across as a grave, intentional misrepresentation at worst, or an intellectual emergency at best. It is completely out of place and carries the tell-tell signs of an echo chamber.
Tiobe has lost all credibility
I have to be honest. I have never taken Tiobe that serious, because they have made to many mistakes in the past to have any form of credibility when it comes to Delphi and Object Pascal as a language. And when I say mistakes, I mean monumental blunders that just annihilate all possibility that they treat languages on equal footing.
“not only have Tiobe failed in their indexing, they have completely and utterly misunderstood the demographic in which the language is used”
If we go back a decade, Tiobe actually based their numbers on the keyword “Pascal”. In other words they excluded not just Delphi commits to GitHub, BitBucket and similar services – they also managed to exclude Freepascal and every subsequent dialect that signify Object Pascal as a whole. So for quite some time their entire statistics was based on the off chance that people typed “Pascal” in their project or commit entries.
To make matters worse, their search tech was not smart enough to recognize “Pascal” in composite words. So if you wrote “ObjectPascal” in a single word, the commit was excluded; As was “Freepascal”, “Smartpascal”, “Oxygenepascal” and variations using a hyphen (and the same for abbreviations).
Developers also use the term Lazarus and FPC interchangeably since Lazarus typically means people use the LCL, the visual framework used to write desktop applications with Freepascal. So while Freepascal has nothing to do with Delphi in terms of intellectual property, the two compilers are used by the community as a whole.
But let’s look at why Tiobe’s indexing fails for Delphi. Just what are they doing wrong?
- Delphi has been around for 25 years, and it’s roots stretch back to the birth of C. Using Stack Overflow as an indicator for popularity is ludacris, since the majority of errors and problems have been largely ironed out in the past, leaving only extremely advanced and rare topics. If problems is the criteria, then I guess that explains why C# and Java soars in the ranking.
- Nobody searches google for “Delphi programming”. You search for explicit topics like composite polygon clipping with GDI+ and then add “delphi” to limit the search to said language. Just like C/C++, Object Pascal is an archetypal language. It stretches from kernel work with inline assembly, to cloud services and HTML5 rendering. So the topics people search for are usually straight out of the operating-system strata.
- Delphi developers communicate in dedicated groups, such as Delphi Developer on Facebook. There is also a thriving community on the Delphi Praxis forums, not to mention the Freepascal forums. None of which seem to be included in Tiobe’s activity statistics.
- Object Pascal has several frameworks and run-time libraries. Delphi ships with two:
- Classical VCL for Windows development (which has seen enormous optimization in the past 8 years, covering UWP, WinRT et-al).
- And then there is FireMonkey (FMX) for multi-platform, embedded and mobile targets.
- Freepascal operates with its own, open-source variation called the LCL
- Freepascal also targets WebAssembly and JavaScript and have variations of the LCL adapted those targets
- And then there is third party, commercial alternatives that covers HTML5/JS like TMS WebCore, Smart Pascal, Oxygene Pascal and the upcoming Quartex Pascal. Around these runtime libraries (VCL, FMX and LCL) there are thousands of libraries, components and frameworks, large and small, that don’t necessarily put “Delphi” or “Object Pascal” in their metadata.
- Tiobe also fails to include feeds like BeginEnd.net or DelphiFeeds, which syndicate on average 3000 unique blog-posts a year, representing a consistent and very much alive stream of information and content.
Delphi and Freepascal, which represents the most widely used compilers, are predominantly used to write commercial, closed source products. Which by consequence means that code and the activity involved is not public. For Tiobe to so utterly misunderstand the demographic for Object Pascal in general, is quite frankly outrageous. If you are going to rank a language that involves millions of users -then at least have the decency of investigating the communities it involves.
Excluding the factors I have outlined above, makes as much sense as excluding mono from C#.
Incompetence or plain ignorance?
It was only after an avalanche of complaints in 2014, orchestrated by yours truly, where members of the Delphi Developer group on Facebook sent complaints en-mass to Tiobe that they addressed the use of “Pascal” to represent Delphi and associated dialects. Yet for all the complaints, outlined in letters that no sentient human being could misunderstand – all Tiobe managed to do was to add “Object Pascal” to their list. Which, believe it or not, was unfamiliar to them.
But do you think they bothered to do it right? Afraid not. Instead of aggregating all of the dialects, frameworks and variations of names under a single banner, they still to this day operate with two very specific search elements, namely “Delphi” or “Object Pascal”.
I sure hope the dairy industry doesn’t hire Tiobe to do statistics on milk, because if their coverage of Object Pascal is anything to go by, they will be ranking by yogurt.
No updates since 2018? Really Tiobe?
When a global Index service like Tiobe manage to write, and I quote:
“However, the latest Delphi release is from 2018” -Source: Tiobe, March report
You really have to ponder if human beings are involved in their business at all. I’m not expecting much, honestly, but I do expect them to interact with the community they supposedly track and build a statistic on. Have they visited Delphi Developer and talked to the admins about growth numbers? Have they talked to Embarcadero to get some figures and coverage there? Did they contact the Freepascal community to get some download statistics from them?
Delphi 10.3 was released on november 21st 2019. The version that Tiobe seem to think is the last update, is in fact the last release with a city name (which was launched in 2018). Since then there have been three successive, regular updates; most developers are now using version 10.3.3. With 10.3.4 about to be released. This just underlines how oblivious Tiobe is to our part of the industry.
Delphi and Freepascal is different in more ways than one, but beyond language compatibility there is one aspect that is quintessential for them both; namely their role in the commercial sector. Where other languages, like C/C++ or (for example) JavaScript see a lot of open-source activity, especially with regards to Linux and Node.js – Delphi and Freepascal are predominantly used to write high-quality, commercial, closed source business applications. In other words, the vast majority of code produced by the millions of Object Pascal developers around the world – is never publicly committed to GitHub or BitBucket.
So not only have Tiobe failed stupendously in their data mining operation, they seem oblivious to the demographic in which the language is used.
Developers use Delphi and Freepascal to deliver rock solid, data driven services; services that is expected to run 24/7 with zero downtime, processing millions of transactions. Delphi is used to write medical software that manages networks of hospitals, with tens of thousands of patients. Delphi is used by banks to power their ATM machines, and Delphi is used to do the heavy lifting in thousands of POS (point of sale) terminals across Europe. Terminals that don’t have time to wait for a garbage collector to kick in, only to cause catastrophic CPU spikes (I won’t mention names, but attempting to switch to C# was a disaster for one of the biggest POS terminal suppliers in Europe).
Object Pascal is used by governments, fortune 500 companies and the guy with a million dollar idea working out of his parents garage; It is used to write cloud accounting software, invoicing systems and medical journaling; It is used by the music industry and graphical design. There are large and extremely successful products out there that don’t advertise that it’s written in Delphi (just like you don’t stamp “made with C++” on a piece of software). You would be surprised!
Object Pascal it’s used by developers who value speed, security, creative freedom and the benefit of a mature feature matrix that only C/C++ and Object Pascal brings. C is by definition three years older than Pascal, but these two archetypal languages have evolved side by side.
There is a reason these two languages represented the university curriculum for close to two decades; further still if we include Turbo Pascal. And Delphi is once again returning home to academia. To the applause of teachers who were forced to teach Java, and hated every minute of it (I helped setup two universities with Delphi in Norway, so I have some first hand accounts in the matter).
Reflections
Since Delphi is growing aggressively these days, Embarcadero is making waves. A few months back we saw how a well known team of C# influencers took a stab at Delphi (and me in particular, no doubt because I have been so outspoken). And as Delphi now returns to academia – Tiobe is demonstrating a bias that leaves little to the imagination. Especially when you know their numbers account for nothing and are bordering on fiction.
If I didn’t know better, I would say someone is worried. And it’s not the Delphi and Freepascal communities. Modern Delphi is a power-house for software development, and it has the potential to disrupt and restore the devtool market.
There is a lot of money involved, so I am not surprised we are seeing a string of attempts at undermining the importance of Object Pascal. I had hoped Tiobe would adopt a higher standard though.
Then again, the ship of credibility sailed when they couldn’t tell Turbo Pascal from Object Pascal.
Leave a comment Cancel reply
Recent
The vatican vault
- September 2023
- August 2023
- March 2023
- February 2023
- December 2022
- October 2022
- January 2022
- October 2021
- March 2021
- November 2020
- September 2020
- July 2020
- June 2020
- April 2020
- March 2020
- February 2020
- January 2020
- November 2019
- October 2019
- September 2019
- August 2019
- July 2019
- June 2019
- May 2019
- April 2019
- March 2019
- February 2019
- January 2019
- December 2018
- November 2018
- October 2018
- September 2018
- August 2018
- July 2018
- June 2018
- May 2018
- April 2018
- March 2018
- February 2018
- January 2018
- December 2017
- November 2017
- October 2017
- August 2017
- July 2017
- June 2017
- May 2017
- April 2017
- March 2017
- February 2017
- January 2017
- December 2016
- November 2016
- October 2016
- September 2016
- August 2016
- July 2016
- June 2016
- May 2016
- April 2016
- March 2016
- January 2016
- December 2015
- November 2015
- October 2015
- September 2015
- August 2015
- June 2015
- May 2015
- April 2015
- March 2015
- February 2015
- January 2015
- December 2014
- November 2014
- October 2014
- September 2014
- August 2014
- July 2014
- June 2014
- May 2014
- April 2014
- March 2014
- February 2014
- January 2014
- December 2013
- November 2013
- October 2013
- September 2013
- August 2013
- July 2013
- June 2013
- May 2013
- February 2013
- August 2012
- June 2012
- May 2012
- April 2012
Try job hunting for Delphi jobs. They exist, but chances are you have to move quite a ways to get to them. As for 10.3, that wasn’t an update so much as a bundling of a 3rd party product that already existed and a few bug fixes. I wouldn’t call anything there “new”.
Delphi development has crawled since they decided to integrate the enterprise…
Since then, Embarcadero has drove away some of the most skilled employees it has, in favour of contracting the work out, it wasn’t picked up the pace.
Sure, we get a few bug fixes and a few api ports, but beyond that?
We shall see.
Frankly, I wish I could get more Delphi jobs, it is my favorite language, but it can be a pretty restrictive ecosystem when it comes to playing with other languages.
I just did. I had 4 job offers in less than a week. I took one in oslo, but could have worked from home. The problem is that people dont use the groups, like Delphi Developer. Thousands of companies using delphi out there, but until EMB ressurects the job pages, we have to use our network.
I agree, Delphi remains a great tool, and TIOBE is not accurate… But what about PYPL (Delphi is ranked 28) and Redmonk ? (not even in the top 20) 😦
All rankings are innacurate because the reality is that codebases and companies are opaque and invisible. However the most likely to be accurate is the job postings list. By that method, Delphi is not top 40.
Warren Postma
Another reason for not growing the number of Delphi programmers is because the product is relatively expensive. If it cost half the price, I’m sure there would be many more interested in this great development platform.
This is interesting, first time I see something like this on Tiobe. Latest release is 2019, they should correct it. But in fact, Delphi is very niche product and it is hard to find Delphi jobs, in my country, almost non-existent, so people go into more mainstream stuff. I guess Delphi will be around the corner as long as Win32/64 and legacy code exists.
I think that TIOBE index is totally inacurte. Let’s speak about the Java Script for example. How this language can be at the 7th place? This language is the only one with a few exceptions that works on the web front.
I think that hackerrank could get us more faithful index:
https://research.hackerrank.com/developer-skills/2020
JS get first place there, and Delphi at the 9th. I think that it more faithful.
Delphi is the great language. I have use it about 20 years, we develop closed source medical visualisation software, millions of lines. And I agree that many sources on Delphi is not open. But this is not a reason to significantly reduce its ratings.
Good point! Thx for the info
I’d also have to add that StackOverflow is under the similar bias. For many years in a row it is not including the Delphi/Pascal into it’s developer survey (top30 languages iirc), even when it’s clear that Delphi is within top20 and was the most often self-reported under “other” in some years.
Indeed. I remember Nick Hodges took them on and got Delphi properly represented there. Sadly the endeavour has been somewhat thwarted by a handful of brogrammers that act as judge, jury and executioner, creating an environment that is quite toxic. I haven’t bothered with that website in years, because they seem more interested in nitpicking individual words – than making sure good solutions reach the audience. But you are quite right, there is a general bias in the population. This no doubt is partly Borlands fault. A lot of people lost a lot of money when Borland went down, and while the pressure from Microsoft is the root cause, Borland did drop the ball on their last releases. So I dont blame people for holding a grudge against Borland, but i do think its strange that they categorically refuse to give the product a second chance. The product never failed. Management did.
Jon, Stack Overflow did include a write-in option at one point. Even with Marco et al whipping up the user base to vote, Delphi couldn’t crack 2% representation. Stack Overflow added any tool that got significant results into its subsequent surveys; Delphi failed to make the cut.
The Stack Overflow data scientists did a lot of data analysis from the survey that allowed Delphi and many interesting things were learned, including that Delphi users’ pay doesn’t remotely match their level of experience. In essence, almost all the Delphi users had decades of experience (abolishing the myth of the new Delphi developer) but they weren’t getting paid in relation to that experience, also abolishing the myth of Delphi being used in the real world much outside of maintaining legacy software.
I’ve debated Nick Hodges and others successfully on this subject before – there is no objective evidence that there ever were “six million” Delphi developers. In fact, all quantifiable evidence points to a very small handful of legacy Delphi developers left in the world. Even Nick Hodges at one point confessed to believing that Embarcadero’s user estimates were off by an order of magnitude.
I find it amusing you’re attacking TIOBE; the Delphi diehards used to dismiss it when it ranked Delphi poorly. When you and Bruce McGee started gaming the system in various ways to bump the figure up, Marco began running blog articles announcing Delphi’s rise in TIOBE! Now that there’s a negative article, it’s “lost all credibility” again. The fact is its methodology is fundamentally flawed, as many commentators have pointed out in the past. It actually OVER-estimates Delphi and older technologies because of its emphasis on search results, even if a page hasn’t been updated or viewed since 1999.
The problem for those who wish to advance the “Delphi is dominant” hypothesis is that all other ranking methodologies that have received backing from data analysts and others paint Delphi in a worse light that TIOBE ever did. Hence, TIOBE is both friend and enemy.
As someone who works in data analysis, I’d love to have a longer discussion with you at some point if you’re open to it. I’ve spent years trying (somewhat in vain) to analyze what I’ve come to call “The Myths Of Delphi” with objective evidence to get a true picture. I admire your… passion, but I believe it’s a lot more helpful to let figures speak for themselves rather than accuse any source that disagrees with your hypothesis as being biased or otherwise working against you.
I’d be glad to share some other figures with you. I did a lot of original research before entities like Stack Overflow opened up their data and made it easy to access. The sad reality is that no objective measure of programming language adoption paints the rosy picture its fiercest advocates promote. But as David Heffernan once wrote, “a language doesn’t have to be popular to be successful”. And as Saul Alinsky observed, you can’t get to where you want to go without first knowing where you are now. If the community could understand (and accept) where they are now, realistic plans could be made for getting to where they’d like to be.
And for April ’20 they’ve now even split Delphi and Object Pascal *facepalm*
officially there is no object pascal at Embarcadero
For the whole Germany there are only 27 jobs listed. But most of them are listed as “Fundierte Kenntnisse in der objektorientierten Programmierung (C#, Java oder Delphi)” so they are not pure Delphi jobs. I counted pure Delphi jobs and there are only 9.
Delphi is long time dead. I really really hope Idera can bring it back. But right NOW Delphi is almost like nonexistent.
Interesting…. some comments that disagree with your view have been censored. I like your attitude. How much is Emba paying you for this article? This makes the reality about Delphi even more sad! I shall post a screenhot about this on my website.
Nice article
delphi is still alive
still alive an kicking
Hope Delphi will comeback in Tiobe Top 5.