Does anybody have any current figures as to the relative popularity of Lazarus on different CPUs and OSes?

I'm watching a discussion elsewhere where people are speculating on whether Delphi will support x64 and/or the Mac in finite time, and knowing what Lazarus users are running could be interesting input.

Mark Morgan Lloyd wrote

Does anybody have any current figures as to the relative popularity of Lazarus on different CPUs and OSes?

Checking the sf download figures might give a rough estimation. Maybe someone can get snapshot downloads because those aren't tainted because those aren't in the official linux distros.

Thanks Florian, I started to think along those lines after I posted. Is that info only available to an admin or should I be able to find it myself (I've had no success so far)?

Assuming that a significant number of Lazarus users do or did use Delphi, the interesting statistic is whether more are developing for the Mac (OS X) or for x64 (Windows/Linux) since that might give some indication of what priorities Embarcadero should have if they really are interested in progressing Delphi as a live product.

I don't know how anybody else feels, but I'd have no objection to a "popularity contest" extension which submitted anonymous platform details when (e.g.) Lazarus or FPC was first installed or had a version upgrade. If it could track LCL usage it could possibly answer some questions like how many people are still using GTK v1.

Mark Morgan Lloyd wrote

I don't know how anybody else feels, but I'd have no objection to a "popularity contest" extension which submitted anonymous platform details

This would be cool! Maybe even FPC itself could create these statistics. Something like this with really big numbers (after aggregating all users) would look quite impressive:

Millions (billions?) of Lines compiled

this month last month total lines total time host arch
total xxxxxx yyyyyy zzzzzzz tttt 386 xxxxx yyyyy zzzzzz ttt x64 xxxxx yyyyy zzzzzz ttt arm xxxxx yyyyy zzzzzz ttt [...]

target arch
total xxxxxx yyyyyy zzzzzzz tttt 386 xxxxx yyyyy zzzzzz ttt x64 xxxxx yyyyy zzzzzz ttt arm xxxxx yyyyy zzzzzz ttt [...]

cross comp.
386->x64 xxxx yyyy zzzz tt 386->arm xxxx yyyy zzzz tt [...]

host OS
win32 xxxxx yyyyy zzzzzz ttt linux xxxxx yyyyy zzzzzz ttt osx xxxxx yyyyy zzzzzz ttt wince xxxx yyyy zzzz tt [...]

target OS
win32 xxxxx yyyyy zzzzzz ttt linux xxxxx yyyyy zzzzzz ttt osx xxxxx yyyyy zzzzzz ttt wince xxxx yyyy zzzzz tt [...]

cross comp.
win32->wince xxxx yyyy zzzz tt win32->linux xx yy zz t win32->osx x y z t [...]

I have to say that I don't know how to write this sort of thing, either from the point of view of Lazarus or from the point of view of making it acceptable to users rather than looking like intrusive spyware.

But I think the really useful thing, far more than just aggregating number of lines of code, would be picking up when somebody tried to compile either Lazarus itself or an app for a specific CPU/OS/widget set, including cases where that combination failed.

How many people get Lazarus (on any platform) and try it out without joining the mailing list or commenting on a project forum? Of these, how many drop it because they have a problem compiling a project? How can we find out where the problem areas are, without getting feedback from these anonymous users?

If we had- for example- aggregated statistics for "This copy of Lazarus" on one of the about tabs, how could we also roll in e.g. somebody attempting to build Lazarus from the command line (make clean bigide) who found it didn't work because he'd got the wrong version of the KDE libraries (been there done that)?

If we had a "check for updates" feature that would give much of the desired information and most users would not see that an intrusive spyware action. Of course such feature is not easy to implement and it had to be really useful, like being able to upgrade "in place".

Paulo Costa

To be brutally honest, I don't like any form of "ET, phone home" things in a program.

Why not?
Because it is nobodies business [how / if / what for] I or anybody else uses the compiler. It has been put out there as open source, no strings attached.

If you want usage information then ask for it on the website or here in the mailing list. Let Lazarus pop up after first install asking for usage info, but never ever have Lazarus/fpc send info in the back ground.

Regards,

Ger

I believe that asking for it on website/mailing list/whatever other way wont help, because people want product which works. They don't want to "lose time" answering questions.

I daily use Eclipse for Java development, and I have usage info turned on to automatically send info in the background. I was clearly asked if I want to do it or not before first action.

So, I believe that implementing such solution would really help fpc/lazarus developers to understand target audience and their needs.

Aleksa Todorovic wrote

So, I believe that implementing such solution would really help fpc/lazarus developers to understand target audience and their needs.

If it's an opt-in (NOT an opt-out) thing then I might be persuaded, but even then I have very big reservations. It's too dang easy to include whatever you may find worth knowing once you down that path.

+100000000000000 ;)

agreed 100000000%

If we are going to send info, can we incorporate a key-logger, and send credit card numbers? So we can find out about the wealth distribution amongst the users
.... only joking ;>

Mark Morgan Lloyd wrote

Does anybody have any current figures as to the relative popularity of Lazarus on different CPUs and OSes?

There you go, I have put the Source Forge statistics in a table:

http://wiki.lazarus.freepascal.org/User:Sekelsenmat#Lazarus_and_Free_Pascal_usage_statistics_from_Source_Forge

The only surprise for me was Nintendo DS. wow, it has a lot of downloads, and then people didn't upgrade much to the latest FPC. It seams that development for game consoles also attracts a huge number of developers, but those consoles only live a couple of years, so when they become obsolete you need to start over.

I'm watching a discussion elsewhere where people are speculating on whether Delphi will support x64 and/or the Mac in finite time, and knowing what Lazarus users are running could be interesting input.

Embarcadero already knows which ports would make the most money (not necessarely the ones with most downloads), and they already started working on both x64 and Mac. The problem seams to be when/if they will be able to release a working product with these features.

How many people get Lazarus (on any platform) and try it out without joining the mailing list or commenting on a project forum? Of these, how many drop it because they have a problem compiling a project? How can we find out where the problem areas are, without getting feedback from these anonymous users?

About getting usage statistics from anonymous users, I don't think we will be doing that, because it would be a maintenance problem to keep a server running to receive that data and then who would read it? Helping people which contribute something (even if their time to ask questions in the forum) already makes at least me busy enough, so that there is no time to help anonymous people. Lazarus/FPC isn't a huge company with 100s of developers and 10os of quality assurances running around all day working only on that.

Note that for many companies anonymous users actually contribute, because often they also pay, so that's why those companies try to support them. Not to mention that a lot of those statistics sending might actually be only psycological. Do you really think that Microsoft reads all the dozens of millions of crash reports that it receives daily? How much time would it take to find reproducible steps for each of them only based in a memory dump and some lines written by the user? Even with real bug reports it is often hard enough to figure out what is going on, and you need further feedback. For an operating system even worse, as often the problem could be related to a specific hardware and so on. How many QAs would they need to do that? 100.000? lol! So don't beliave too much on that data sending ;)

SF is only a small part of FPC's downloads. The fpc master ftp has approx. 25,000 downloads (less in summer) of fpc-2.4.0.i386-win32.exe per month. So approx. 300,000 per year and this is only one mirror. So the reason for the high Nintendo DS download number might be that some webpage/blog/wiki with a FPC Nintendo DS howto links directly to the SF page and thus the statistics gets wrong.

About the Mac OS X "market": well, fpc-2.4.0.intel-macosx.dmg has 300 to 400 downloads per months.

Numbers can be found at: http://www.hu.freepascal.org/webalizer

Ger Remmers wrote

To be brutally honest, I don't like any form of "ET, phone home" things in a program.

I agree, and I also don't like mandatory registration.

Why not? Because it is nobodies business [how / if / what for] I or anybody else uses the compiler. It has been put out there as open source, no strings attached.

Very true, but it /is/ in your interest to make sure that the developers are aware, in some structured way, what facilities you're using.

If you say what widget set etc. you're using in the ML then it might be overlooked, there's definitely no easy way to mine info like that. If you put it on a bug report then it's possible to mine info from people who've reported problems but not from those who haven't. And so on.

Paulo Costa wrote

If we had a "check for updates" feature that would give much of the desired information and most users would not see that an intrusive spyware action.

I wonder what an update is in this context? How does one ask svn whether anything "worthwhile" has gone into the fixes branch?

Of course such feature is not easy to implement and it had to be really useful, like being able to upgrade "in place".

It would probably be possible to add a "rebuild when next terminating" facility. More tricky would be to make sure that everything was rebuilt- all widget set interfaces, cross-compilation variants and so on.

Ger Remmers wrote

If it's an opt-in (NOT an opt-out) thing then I might be persuaded, but even then I have very big reservations. It's too dang easy to include whatever you may find worth knowing once you down that path.

Yes, I agree. But something like this doesn't need to extract serialisation etc. from the computer, it would be desirable to ignore multiple reports from the same instance of Lazarus where the configuration hadn't been changed but this could be identified by a timestamp of when it was first fired up.

I agree with this. But we've to life with its downsides: We've no clue in what directions the compiler should be pushed. The mailing or bug lists are a bad indication for this: just imagine 90% of the users use a certain compiler package but never have problem with it, so no bug reports and no mailing list posts. So we decide, ok, let's dump this particular release type/package, it seems nobody is interested in it.

It boils down to: you usually hear only form unsatisfied customers and seldomly from the satisfied.

It has been put out there as open source, no strings attached.

Actually free software ;)

Many thanks for that Felipe, I'll pass it on. If I'm reading that correctly there's going to be a few additional x64 users hidden in the Linux bucket, so the number of x64 (all OS) and MacOS (all hardware) users are probably fairly similar.

Games consoles and phones are something I've never really looked at, although occasionally somebody asks me whether they'd be any use as the brain of an embedded system.

But would the ports that made the most money in the short term be the ones to keep them in business in the longer term? If for example they predicted lots of NintendoDS sales they'd get bitten when the platform became obsolete, and they could get similarly bitten if Apple continues to spread FUD about the degree to which "alien" languages are welcome on their products.

About getting usage statistics from anonymous users, I don't think we will be doing that, because it would be a maintenance problem to keep a server running to receive that data and then who would read it?

Very true. But it's not so much who would read the raw data but whether it's possible to get reliable information on e.g. how many people are still using GTK1 when the developers decide to axe support.

If it were the developers probably wouldn't do as fine a job :-)

Do you really think that Microsoft reads all the dozens of millions of crash reports that it receives daily?

Memories of reading coredumps on music-ruled paper. On the other hand I'm currently chasing an intermittent termination problem in live code on a handful of application servers: having a million users file automatic reports when it happened would save me having to wait until the end of the year to see whether my fix has worked.

I understand what you are saying, but no, it's none of Florians business (sorry Florian) to know what I'm using, or how I''m using it.

If you say what widget set etc. you're using in the ML then it might be overlooked, there's definitely no easy way to mine info like that. If you put it on a bug report then it's possible to mine info from people who've reported problems but not from those who haven't. And so on.

I fully understand the dilemma, but to me it's my interests first and screw the rest. If I have something to [whine / bitch / complain] about I will do so in here

Yes, and the same way I usually don't care what users want (use it or use it not), but I develop what I want :)

I understand the frustration, but still..... No "ET, phone home" thingy without approval of the user........ ever. And even then you'll have to be very careful about what you want the system to report back.

Just keep in mind that MS has tried this a few times and every time they got burned over it.

> It has been put out there as open source, no strings attached. Actually free software ;)

Okay, agreed, but still.......

No "ET, phone home" thingy without approval of the user........ ever. And even then you'll have to be very careful about what you want the system to report back.

As explained, for me there is no need for such a thing, I wouldn't listen to it anyways. The same way as I don't listen to teasers like "if you do this or that, fpc will get really usefull and great or whatever".

And that's the way I like it best. You do what you want and I do what I want and if we meet somewhere in the middle, fine. If we don't... so be it.

I think this renders this argumentation kind of useless then :)

Ger Remmers wrote

I understand the frustration, but still..... No "ET, phone home" thingy without approval of the user........ ever. And even then you'll have to be very careful about what you want the system to report back.

Nobody's arguing against that.

Ger Remmers wrote

I understand what you are saying, but no, it's none of Florians business (sorry Florian) to know what I'm using, or how I''m using it.

Actually I don't think that's true. I don't think that FPC has been placed in the public domain so Florian (or somebody) still retains nominal ownership and copyright. As such he emphatically DOES have a right to know that you're using the compiler and its related subprojects in accordance with the licensing terms (LGPL etc.), and that's probably particularly important in the case of e.g. the TurboPower stuff.

What is more I think that he has that "right to know" irrespective of your views or consent. Whether I like that or not is irrelevant.

I'm sure that you're aware that there is a small but growing number of companies who have been forced to admit that they were abusing the (L)GPL: do you think that they consented to the software's owners finding out that they were doing that?

Depends on whether you're looking at it from the point of view of the underlying compiler or of the IDE and class library which is built on top of it. They're related technically, but they're still separate projects.

Besides which, I don't think that anybody's taking a particularly strong position that it's something worth putting time into. So there's not much of an argument.

Of course, the source code remains property of the developers.

And here we disagree. No developer has "the right" to know what I use the program for as long as I don't claim the program to be my product.

I wasn't aware of that ( I don't keep tabs on the software world).

At first sight that appears to be a valid position. However considering the case below Westinghouse weren't claiming that they owned BusyBox, merely distributing it with complete disregard for the license. Hence the owner /does/ have a right to know it's being used, even if the user is entirely silent about what they're doing with the software.

The fact that in some cases the owner goes on record as saying that he's not currently interested doesn't mean that he doesn't have the right to know.

http://www.theregister.co.uk/2010/08/04/gpl_violation_westinghouse/

With respect, if you aren't aware of what's going on you might refrain from commenting on who's got rights to do what. Anyway, this probably isn't a good place to continue this subthread.

2010/8/11 Ger Remmers

I wouldn't mind if on the first run of the Lazarus IDE after a new install, the IDE asked me if it could send some info to the web, like FPC version, OS, CPU, LCL version and Widgetset.
Jo?o Marcelo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lazarus.freepascal.org/pipermail/lazarus/attachments/20100812/a74f8c5c/attachment.html>

Not wanting to disclose what one uses a program for does not equal disregard of the license.

Not knowing who gets robbed by whom doesn't make it that I wouldn't know robbing is against the law.

Anyway, this probably isn't a good place to continue this subthread.

What place is better?

I wouldn't mind if on the first run of the Lazarus IDE after a new install, the IDE asked me if it could send some info to the web, like FPC version, OS, CPU, LCL version?and Widgetset.

Depending on whether there is anonymity in the information. Anyway would not be very nice.

Ant?nio

On LazarusBrasil I have this page (
http://lazarusbrasil.org/pesquisa.php ) to get user information. Few answers, but nobody can question its ethics.

Ant?nio

Perhaps it would be better to bet on the user's commitment to the community and simply ask him to respond.

Ant?nio

Maybe the software owner has a questionable "right to know" about what is done with his product, but the user also has the "right not to tell."

Ant?nio

Ant?nio wrote

Perhaps it would be better to bet on the user's commitment to the community and simply ask him to respond.

That sort of thing is fine provided that everybody knows about it and that it's not abused by people responding multiple times to push their private agenda.

That sort of thing is fine provided that everybody knows about it and that it's not abused by people responding multiple times to push their private agenda.

Pushing private agenda is a right of the user. You should know how do you interpret the answers.

Ant?nio

True, but no license requires me to disclose what I use the software for.

Nowhere in the lgpl does it state that the user is required to inform the developer of any and all uses of said software.

Every user needs a license for using a software. When that license requires a registration (or the like), then this requirement is binding.

DoDi

Maybe the software owner has a questionable "right to know" about what is done with his product, but the user also has the "right not to tell."

Everybody has the "right not to use a software". But when he wants to use it, then he has to obey the license agreement.

DoDi

True, but the license is the author's decicions and an author could write a license which requires to submit usage info. If one doesn't like it, don't use it.

True, but the license is the author's decicions and an author could write a license which requires to submit usage info. If one doesn't like it, don't use it.

At least according to the laws of most Western countries, no contract may contravene the law and it recognizes the right to privacy.

Very true, but I think that Ant?nio was thinking of the general case where the owner of a software project has the undisputed right to specify what sort of license(s) he is going to release it under. In some cases he says that he wants a picture postcard from each user, and if that's what he says then it's /not/ negotiable.

Anyway, I was responsible for starting this thread and I'd suggest that any further general discussion of owners rights and various licensing conditions would be much better off in e.g. the lazarus-other list. For anybody who missed the announcement the subscription page is at http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus-other

Correct, but I wasn't referring to some hypothetical what-if case, but to the actual used gpl licenses in their various forms in the Fpc/Lazarus set.

Well, even with the gpl, I see nothing in it which forbids it?