Why isn't Lazarus / Free Pascal more popular?

[Lenny33]

Have you ever asked your mechanic which screwdriver or socket wrench he likes best?

I can play alegories too.
By the way, I still have my grandfather's wrench in my shed in a place of honor. It's about 90 years old. It has Ford markings on it. My grandfather drove an ancient Ford in the 1930s.
But I don't do anything with this wrench anymore. I have modern wrenches and fast and powerful cordless impact wrench.
I hope you understand the alegory.
FPC/Delphi every year turns more and more into this honorable wrench on the shelf, which almost nobody (mostly except old farts) wants to use anymore.
It takes some serious effort to change that. And at least to start with recognizing the real problems that do exist.

[LV]

FPC/Delphi every year turns more and more into this honorable wrench on the shelf, which almost nobody (mostly except old farts) wants to use anymore.

Something similar can be found on the Rust forum. 

C++ is becoming more and more like a gun for shooting yourself in the foot, which almost no one (except maybe old farts) wants to use anymore.

[Handoko]

Out of curiosity, I googled out and got this:
(source: https://www.i-programmer.info/news/184-cc/13986-c-picked-out-by-tiobe-an-odd-choice.html)

[d2010]

Hi,
Despite all of those facts Free Pascal seems to be a dying language. Most of the devs seem to use it for many years, working on legacy software, which makes me think that the majority of devs is a bit older.

The Visual Basic, or C++ or Go,  they scan more deeply inside recursive algorithms,
But delphi and lazarus, do not scan more deeply recursive algorithms.
Anyone ,, one solution for "future.Lazarus" for "lazarus" is the answer of this trick
Emulate of "scan more deeply inside recursive algorithms", and follow the next
question.

Code: [Select]

Eu multumesc ArsenieBoca, 2025.
Eu multumesc ArsenieBoca, 2025.
C:Q1=Can "Lazarus.exe" have emulator for "scan more deeply inside recursive algorithms"?

Code: [Select]

Eg1: "Lazarus.exe" have emulator  using "String" for others dates-type and not using widechar or not ansistring;, because old-dos-string have good speed inside "inside recursive algorithms". Please, please  Your team (c)Lazarus do not use widechar or ansi-string, or
not  GetMem(Buffer), Free, ReallocMem(binary), FreeMem(binary),, these function
crack/slow the emulator.


[egsuh]

Is it fair to compare languages like C or Pascal with javascript, R, or Python? Might be compared with Python, but javascript and R are for different purposes, I think.

[marcov]

except old farts

Apparently it seems young farts are more susceptible to pointless popularity contests.

P.s. if your monte carlo example uses system.random, the comparison is flawed. I couldn't quickly find the sourcecode for the delphi binary to check myself.

[Lenny33]

Apparently it seems young farts are more susceptible to pointless popularity contests.

It's simple. Young farts look where there's more potential and money-making opportunities

P.s. if your monte carlo example uses system.random, the comparison is flawed. I couldn't quickly find the sourcecode for the delphi binary to check myself.

No, the same simple C-like function is used for all sources. But VC++ optimizes it well and makes it inline despite its size.

[silvercoder70]

Lenny33 said ...

FPC/Delphi every year turns more and more into this honorable wrench on the shelf, which almost nobody (mostly except old farts) wants to use anymore.

At what point in time is a person considered an "old fart"?

I don't have that many subs on my Youtube channel. That said ... Less than 30% of the subs (according to YouTube) are 55 and order. The rest are between 18 and 54. While about a quarter are in the 40+ age bracket, there is approx. 20% in the 25-34 age bracket. Which sounds OK to me.

Stats, stats, and more stats! 

[Nimbus]

Every tool has its use. Pure python would be orders of magnitude slower in those tight loop benchmarks than compiled languages, but that does not mean python should go to the shelf in any way - yet many young people choose it, cause it's good enough for what it typically does. Java would beat FPC, (and in some circumstances, even C++) because of its JIT, but you wouldn't want Java in your time-critical signal processing because of random freezes when the GC kicks in. C++ is faster at runtime (but often horribly slow while compiling) yet leaves more space for human errors in development, so there are always trade-offs. There's also the startup times and scalability factors, which these type of benchmarks do not account for.

So it all depends, the decline in popularity of Pascal is certainly not in that the "bad" compilers produce "slow" code, it's a bit more complex.

[440bx]

Stats, stats, and more stats!

Some "stats" can be found at:
https://forum.lazarus.freepascal.org/index.php/topic,61276.msg460570.html#msg460570

Admittedly, the sample is small but, it's likely favorable to Pascal/FPC because it is comprised of forum members.

The OP's question reminds me of something Jerry Pournelle stated in one of his articles in (the equally defunct) BYTE magazine, it was that the success of a programming language is determined by the percentage of hard-core programmers who use it.  (certainly true for C and C++.)

At least in the case of Python, the really good stuff is hard core libraries written mostly in C/C++ which enables the average programmer who couldn't dream of doing some things if those libraries didn't exist.  A variation of that applies to Java.  The portability these language enjoy is also a significant bonus in more ways than one (a programmer well versed in one of those language in one platform can somewhat quickly become productive on another platform.)

When it comes to UI components, I remember when I was looking at C#, just about anything I could dream of as far as interface widget was available or easily obtainable by making small modifications to an existing widget.  That's what Lazarus is competing with, the .net framework and all the gizmos it provides the programmer.

There are other critical reasons and there is no indication they are going to change either.
 
When comparing Pascal to C/C++, performance is important but, it is far from the only reason C/C++ are more popular.

[marcov]

Apparently it seems young farts are more susceptible to pointless popularity contests.

It's simple. Young farts look where there's more potential and money-making opportunities

That is what they tell themselves, yes. 

Seriously, I haven't bothered since I have heard this stuff all before in 25 years+ Delphi and Lazarus. Instead I'll list some experiences from me over the years:

• Talking only about language is overrated. You choose a complete development target (compiler W for language X with libraries Y and widgetset V on target Z, not just language.
• Speed isn't everything. and most applications are not one source file math benchmarks anyway. Do your own benchmarking, but directly related to your application domain and tradeoffs and mostly ignore the generic public ones. That is only discussion fodder. Model for your application (checked into detail), or don't bother with benchmarks till you have actual problems
• I work in image analysis (with Delphi), and have 1000 lines of assembler and use C++ simdlib for a few special cases (simdlib is not used in production we only used it to prepare some tenders) on half a million lines total  Cases where you have a single rate determining step (like whole image transformation) in a relative few lines, it is usually not that important to factor in doing that in the core language
• When I went pro, I myself back then had the choice between Delphi and VS6. MFC didn't really look elegant and attractive for gui, actually nothing about VS6 was attractive (back on C++ standards etc) though familiarity with Pascal certainly played a part.
• Avoid being responsible for picking the language/tool in mid or larger companies, as suddenly your colleagues and manager blame everything that goes wrong (or is one character extra typing in some random scenario)  on your choice of the tool/language, and you have to constantly defend and walk the extra mile "fix" it. Popularity is a defence in such cases, but it is not always the perfect defence as the criticism might be on a detail.   

[marcov]

Every tool has its use. Pure python would be orders of magnitude slower in those tight loop benchmarks than compiled languages, but that does not mean python should go to the shelf in any way - yet many young people choose it, cause it's good enough for what it typically does. Java would beat FPC, (and in some circumstances, even C++) because of its JIT, but you wouldn't want Java in your time-critical signal processing because of random freezes when the GC kicks in. C++ is faster at runtime (but often horribly slow while compiling) yet leaves more space for human errors in development, so there are always trade-offs. There's also the startup times and scalability factors, which these type of benchmarks do not account for.

So it all depends, the decline in popularity of Pascal is certainly not in that the "bad" compilers produce "slow" code, it's a bit more complex.

I agree mostly with what you say. Though while it has been a while, Java only beat FPC in benchmarks and other very localized batch problems, not in complete programs.

[Martin_fr]

Why isn't Lazarus / Free Pascal more popular?

Well, we had very interesting answers to this over the past decades.

Most of them: "build it and they will come"

1) (Long ago): Because Lazarus is version 0.9.xxx, make it version 1.0.
We were promised and guaranteed that when we would change the version number (and only the version number / functionality and quality were said to be plenty) then new users would come and be plentiful.

2 & 3)  Add Darkmode / Add Anchordocking
Same promise.
Both exist (as addon) for a long time.

Well Anchordocking is build in with 4.0 => so lets see, if those promised new users will show up this time.

4) AI.
Lately more and more hints are coming in that adding AI will do the trick. If only, and nothing else but only AI would be integrated, then we should expect big flocks of new users. (Oh, and add-on for 4.0 exists)


I might have missed/forgotten a few of the "Request & promise" features that have been brought forward during the years.  Maybe "inline var declarations", and "anonymous functions" ... (the latter is in FPC 3.3.1)

Speed, strangely has IIRC only been on the "absent complaints" list, but hasn't come with that promise. Which IHMO at least is a plus point towards the honesty of those complaining about it.

Well, Pascal/Lazarus are not only still here. They are standing their ground, and even growing a bit again.

And given, that there is enough interest in this thread's question, maybe that question should not be: "Why isn't Lazarus / Free Pascal more popular?".

Maybe it should be:

Why is Lazarus / Free Pascal still that popular?

[Khrys]

Pascal itself is perceived as old & outdated. And with the abundance of alternatives nowadays, only few give it the benefit of the doubt.
Modern Pascal may support classes, generics, operator overloading, automatic reference counting and the like, but it seems that nothing can overcome historical inertia (C, C++) and backing by massive corporations (C#, Go).

I see no way to overcome this, unfortunately.

[Fred vS]

I see no way to overcome this, unfortunately.

I see one way: create nice, powerfull, out-of-the-box, multi-platform applications and share it to show what fpc/Pascal can do.