Would you support the development of a FPC GUI frontend?

[Robert Kosek]

My intent certainly is not to start a flame war. I asked because I wanted to know. Almindor and marcov were the ones dropping flamebait, not me. Thanks for correctly reading me, ?ëu?±o. (And everyone else who hasn't gone overboard on this too.)

William, I never said that I didn't like OSS. It has its place, but I cannot be involved when I cannot financially support myself through it. If you haven't read, which I feel you haven't, I have echoed this message somewhere around 5-6 times between the two threads. I have made no plans, no real prototypes, and have only asked a casual question of whether people would support such a project -- the proposed specs are wide. All I wanted to know is if it would be worth the effort it would take to make it, because I cannot do it for gratis/free.

What I hate is the attitude that I have to take what I dislike silently, with no complaints, and the ones who made it aren't responsible for evaluating my position and perhaps fixing what might be broken functionality. There are two issues I haven't reported because of this attitude. I possess no blind loyalty to any compiler, team, or even Pascal itself.

dmantoine, yeah I am aware there are a number of FPC IDE's out there and that I'd have to really compete to make it worth the money. Part of why I was asking this whole question in the first place. I am well aware that it would take some work, but I also can't quite find what I'm after for a FPC IDE currently. So I figured that if I was going to consider putting the work in, at least see if I could sell a license of it, etc. Not to mention a few of the feature requests are out of my range, like built in debuggers -- though given time maybe I could figure it out.

[dmantione]

What I want to make clear with these messages is that there already exist IDE's and that you have to position your project to attract interrest. I.e. I'm sure a project having features like Nu?±o mentions will attract attention.

A duplicatie of say, FPS, won't attract any attention.

Despite this, the question is how to get those features. I.e. we are well aware that debugging is the weak point of FPC currently, especially under Lazarus. This is a direct choice of using GDB, and interfacing with it.

The textmode IDE, includes GDB in its own exe, which has the advantage that it has more influence over debugging, and therefore provides a slightly better experience. However, as only 1 gdb can be present in an exe, it means you need 1 IDE per processor type. I.e. Nu?±o's wishlist of having an IDE that supports multiple CPU's and debugs better than Lazarus, may be incompatible with GDB.

In the end, no GDB based solution can provide a good user experience, simply because gdb doesn't support the Pascal language well enough. So ultimately to tackle the debugging problem, someone will have to write a multi-platform Pascal debugger.

In short, identifying the weak points is one think, solving them is another matter. A new IDE can, by starting from different design decisions provide a different user experience. The LCL is a thick abstraction layer, you may want to decide for a smaller one, with the advantages and disadvantages. But you most definately will run in some of the same issues other IDEs run into.

Don't let that prevent you from writing an IDE. All IDE's started because someone wasn't happy with the existing ones. But do know what you are starting to work on, and in what way you think you can improve on the existing ones. Just saying "Lazarus is crap" and starting your own one without having considered what you want to make and how, is probably not a recipe for success.

[Robert Kosek]

Well, I haven't said that Lazarus is crap yet, but I have said that every time I try and give it a chance something goes wrong or doesn't work. An example would be the StringGrid I was working with the other day, only to discover that TStringGrid.Cells[x,y] always returns a blank string and cannot set the contents; and the EndEdit event doesn't fire.

My primary intent would be an advanced ConText-like editor tailored specifically for FreePascal, with snippets, code completion (eventually smart code completion), the ability to run the standard FreePascal compiler through either commandline directives or automatic makefile generation, and the ability to manage products similar to Lazarus. However it would be a graphical "faceplate" to the existing FPC. I can't claim to have considered or planned any more than that because I haven't. That's all I thought of.

[dmantione]

Okay, understood, but ....., what would the usability improvements be over for example the textmode IDE or Lazarus? Both already do have an advanced text editor and allow you to compile it.

Putting it in a different wording: You'd like people to pay for your program. What improvement in their coding experience are you going to offer them, which makes it worth for people to spend money on?

[JernejL]

What does the ide for fpc need to have:

- Integrated Debugging

I think that's the first thing one should do in an IDE, without that it's pretty useless, if you look at the state of debugging in lazarus you'll see what i mean.

[Mirage]

Fully-featured debugger is a hard to make thing.
I see two possible debugging solutions for FPC:
1. Use Borland's one. They have a very nice Pascal supporting debugger, but I don't know what they think about using it by Free Pascal.
2. Use virtual machine: with integrated debugging turned on the code compiles into some kind of p-code which will be executed by VM. It's an easy task to create an absolutly crossplatform debugger for VM. But make FPC to compile into p-code may be hard. Also executing p-code is slow.

[dmantione]

A VM can be used to make something like a platform independend debuggger work, but won't provide an actual solution regarding user debugging experience since it is very hard to make things like assembler code, pointer tricks, or OpenGL access work in it. I.e. programs cannot be debugged the way they are written.

Using the Delphi debugger would be challenging as well:
- Borland copyright, why would they give it away?
- No support for Dwarf debuginfo
- Only works on Win32
- Has never been used outside the Delphi IDE is probably closely integrated with it

What is needed is a multi-platform capable debugger that supports the Pascal language well. As soon as it exists, IDEs can take advantage of it.

[arthurprs]

A VM can be used to make something like a platform independend debuggger work, but won't provide an actual solution regarding user debugging experience since it is very hard to make things like assembler code, pointer tricks, or OpenGL access work in it. I.e. programs cannot be debugged the way they are written.

Using the Delphi debugger would be challenging as well:
- Borland copyright, why would they give it away?
- No support for Dwarf debuginfo
- Only works on Win32
- Has never been used outside the Delphi IDE is probably closely integrated with it

What is needed is a multi-platform capable debugger that supports the Pascal language well. As soon as it exists, IDEs can take advantage of it.

btw it's the best debugger i ever seen.

[WILL]

Hey guys. I'm just jumping into the conversation half way so please pardon me if I've missed a beat or two of the flow of the conversation.

Here is an idea that I'd like to put out there...

Why not work on this new IDE with the focus on greater debugging capabilities than most of the others out there? On top of that, allow it to be taken separate and reused with other IDE's. This might work best for all concerned in the following ways;

:arrow: Other IDE tools could take advantage of a more Pascal-centric, easy to integrate debugger.

:arrow: The debugger it's self would generate a greater user-base and amount of required feedback to improve development and debugging(ironically enough )

:arrow: It in effect increases the applicability of the Pascal language as a whole.

[marmin]

Suggestion , why not join forces with the Lazarus team, they need it.