Maybe the openness only goes as far as letting people see what they do, and hope that suggestions for improvements (in the form of patches and code) will pour in. A common approach these days for big companies that 'go open source' :)

I would like to add a plea here:

It's not because Borland now made CLX available, that we should all of a sudden jump on the Kylix BandWagon. The LCL has come a long way, and I think that the focus should still be on getting the IDE to a workable state.

Adding support for CLX and other things means in this case: Make sure that lazarus can use both the LCL class library and the CLX; This means that a certain independent-ness of the used toolkit is needed, but since the lowest common denominator is the Classes unit, this should not be hard.

By the time the Lazarus team gets the IDE to a point where it can be used easily, the FPC core team should have made some progress in getting some fundamentally missing concepts such as Variants, Multithreading and widestrings to work in the compiler.

Without support for these things, trying to get CLX to run with Lazarus (or FPC) is a waste of time.

Another thing is that the CLX still contains some pretty basic functionality failures, which I hope Borland will fix soon.

In short: The LCL and IDE should still be the primary goal for Lazarus.

Michael.

That's not true at all... MIDAS isn't required..... and I do think you get the DB libs.... you just don't get the drivers for some things like Oracle. They're not part of CLX.

So would they, and I'm sure that's why it's free.

Simply put: All needed libraries to make it useful.

- No C wrapper library sources for Qt.
So forget about VisualCLX unless you own Kylix.

- No DBExpress, no Midas libraries. The useful database units that are provided _NEED_ these libraries.

So no DataCLX, unless you own Kylix, or you use 3rd party components, of which there are none at this time.

- All NetCLX components are missing altogether.
So no apache or server development, unless you use Indy.

Not to mention that:

- The system unit is only useful for Kylix, because you need special compiler magic to make it work. Which compiler has this ? Kylix only.

- Exception unwinder is useless without the needed compiler support. The only compiler that can work with this is... Kylix.

Michael.

You don't have to believe me, just check the sources. TClientDataset does _NOT_ work without midas.

Have a look at the following (randomly selected) code snippet:

----------------------------------------------------
function TCustomClientDataSet.GetXMLData: string;
var
  DataPacket: TDataPacket;
  VarPacket: OleVariant;
begin
  if Active then
  begin
    CheckBrowseMode;
    FDSBase.SetProp(dspropXML_STREAMMODE, xmlON);
    Check(FDSBase.StreamDS(DataPacket));
    DataPacketToVariant(DataPacket, VarPacket);
    Result := VariantArrayToString(VarPacket);
  end;
end;
-----------------------------------------------------

FDSBase is a object (actually, an interface) that
resides in the midas library.

and I do think you get the DB libs....

No, one doesn't.

I have the cources checked out, and no DBExpress sources, no midas sources, no Qt wrapper library sources.

Unless they put them in another part of sourceforge of course, in which case I will be more than happy to rest my case.

you just don't get the drivers for some things like

But why isn't it in CVS then ?

Michael.

> - All NetCLX components are missing altogether. > So no apache or server development, unless you use Indy. > Indy is free.

ICS and Indy were free already, so we don't need an open CLX for that

We could have reused it. I think Borland is afraid we (or others with our sources) might be able to create a LGPL library for Kylix.

That would force them to pull the GPL version of Kylix, because creating commercial apps with free-Kylix without using Kylix libs might be legal.

Not gonna argue this one...

- No DBExpress, no Midas libraries. The useful database units that are provided _NEED_ these libraries.

You still will have clientdata set and whatnot... really, other than DB drivers, which you *should* pay for, what else do you need?

- All NetCLX components are missing altogether. So no apache or server development, unless you use Indy.

Indy is free.

Not to mention that: - The system unit is only useful for Kylix, because you need special compiler magic to make it work. Which compiler has this ? Kylix only.

Does this matter? Wont' lazarus provide it's own equiv?

Michael Van Canneyt wrote

Another thing is that the CLX still contains some pretty basic functionality failures, which I hope Borland will fix soon.

Out of curiosity, what is CLX missing?

It's funny, under environment it says win32,

Under OS it says Linux and WinNT/2000

I guess win98 users don't rate.

Chris

You still will have clientdata set and whatnot...

Of course, but they don't work without Midas. The clientdataset is just a TDataset wrapper around the Midas lib. Without midas library, no TClientdataset, no MyBase, nothing.

What you HAVE is TDataset, and this is an abstract class, so it needs descendents.

The two usable descendents that come with Kylix, notably: TSQLDataset (and some variations) and TClientDataset needs libraries that are not available.

-> you end up without any DB support at all.

really, other than DB drivers, which you *should* pay for, what else do you need?

Well, the libraries to make it work. The driver is no good without the dbexpress library. Which you don't get...

And I hope you'll agree that paying for a Mysql driver when Mysql itself is free would be strange :)

Yes, but Apache module development is out, as well as CGI development. You can't do those with Indy. (not without re-implementing NetCLX, that is)

Yes. All I wanted to say with that is that making the system unit publicly available is pretty useless, since it'll only work with Kylix anyway.

That said: I've worked with Kylix for some months, and I've seen it improve a lot. Don't think I am against it, far from it, I even wrote an article about it which earned the publisher some congratulations from Borland itself, so...

...but I am a bit sceptical about the 'open source'-ness of Kylix.

(and about some other things about Kylix as well, but that is another story)

Michael.

Greetings,

Borland has given us a heads up about FreeCLX. They have placed the current FreeCLX source on SourceForge and have let us know so that we can start looking at it with regards to making Lazarus work with FreeCLX.

However, the initial evaluation by the FPC team isn't that good. Many of what might be considered essential parts are missing. Other parts that are included will only work with Kylix. I have conveyed some of these concerns to M. Swindell at Borland and we will see what the response is. However the open source'ness and free'ness of CLX that Borland is proclaiming might not be as open as one would hope.

An announcement from Borland about FreeCLX should be forth coming in a few days.

If you would also like to look at the code you can get it out of CVS. Here is the SourceForge location for the project.

http://www.sourceforge.net/projects/freeclx

Oh I don't think any of use consider abandoning the LCL. I believe that we would like to make sure that the Lazarus IDE can be built with FreeCLX just so that it gives the delphi community that tool as an alternative to using Kylix. The one negative with FreeCLX no matter how you look at it is that you can not build commercial apps unless you actually buy Kylix. I myself have a commercial app I want to make available for Linux but will not be able to afford the price of Kylix to do so. My choice is to use FPC and Lazarus. The other advantage is that it will then be portable to a larger audience then Kylix will be able to anytime soon.

Adding support for CLX and other things means in this case: Make sure that lazarus can use both the LCL class library and the CLX;

I don't know that we intend on integrating FreeCLX into LCL in any way. I believe it is just to make sure that the IDE can be built with FreeCLX which if it uses standard unit names and component names shouldn't be a problem. Then it can be used to actually build FreeCLX apps. by either compiling with FPC or calling the commandline version of Kylix.

In short: The LCL and IDE should still be the primary goal for Lazarus.

It is, it is. :-)