BBC BASIC for Windows
Developers' Area >> Future developments >> The future of BBC BASIC for Windows
http://bb4w.conforums.com/index.cgi?board=developments&action=display&num=1330539990
The future of BBC BASIC for Windows
Post by admin on Feb 29th, 2012, 5:26pm

Long message, sorry.

Let me kick off this discussion with some background, so you all know where I am coming from. A few years ago I announced that BB4W was transitioning from a status of 'active development' to that of 'maintenance', in other words new releases would be for the purpose of bug-fixes and other minor tweaks only, rather than major additions to functionality.

There were a number of reasons for that decision, but primarily it was a risk/benefit judgement based on a gut feeling that the benefits of any major additions could be more than offset by the risks of introducing bugs. One of the key characterisitics of BB4W, IMHO, is its stability and freedom from bugs and I didn't want to damage that.

Nevertheless, it wasn't a decision taken lightly. I wanted to reassure myself that there were no serious shortcomings in BB4W that could cause users (or me) annoyance if they were not addressed. I decided that a good way of assessing the 'completeness' of BB4W would be to attempt to emulate another major interpreted BASIC - QBASIC. QBASIC is a little younger than BBC BASIC but is similar in respect of the facilities it provides; if BB4W could emulate QBASIC fairly comprehensively then there ought not to be any serious weaknesses.

So it was that I developed QB2BBC, the QBASIC to BBC BASIC translator. I realised from the outset that two fairly major enhancements to BB4W would be needed: support for labels and the EXIT keyword; neither of those features could have been emulated in an acceptable way otherwise. They were added in BB4W version 5.60a (May 2007), although the support for labels was initially kept 'secret' for fear of the reaction from die-hard BBC BASIC enthusiasts who consider labels the invention of the Devil! Thus the value of QB2BBC in identifying shortcomings in BBC BASIC was proven.

Fast forwarding four years, I decided in 2011 to write a Liberty BASIC to BBC BASIC translator. Since Liberty BASIC is a more modern dialect (originating in the early 1990s) it might prove an even better test of the 'completeness' of BB4W. I was pleasantly surprised at how few tweaks to BB4W proved to be necessary to emulate LB fairly comprehensively: a change to the way strings are passed to SYS and a couple of file-reading enhancements: GET$# TO and GET$# BY, which were implemented in BB4W v5.93a (needless to say I didn't own up at the time that they had been added specifically to support LBB!).

There are a couple of features of Liberty BASIC that BBC BASIC cannot emulate: huge strings (up to 16 Mbytes or more) and arbitrary-precision integers. Supporting either of those would have been far too challenging, and they remain the most significant compatibility issues as far as LBB is concerned. Nevertheless, writing QB2BBC and LBB has convinced me that BBC BASIC for Windows has no serious shortcomings, and that there is no reason to re-visit my decision to discontinue active development.

Therefore, and most importantly, I do not consider modifications to BB4W to be on the agenda. If I release any further updates, and that isn't certain, they will be only to address bugs (such as the one identified yesterday - grrr!) and to provide the opportunity to update example programs and libraries.

And that brings me to what I really want to discuss (sorry it's taken so long to get to the point): how adequate are the existing libraries supplied with BB4W? Because BBC BASIC, as a dialect, pre-dates the Graphical User Interface concepts that are so important these days, GUI support necessarily requires the use of libraries (principally the WINLIBx libraries). They are supposed to make accessing the Windows API fairly straightforward, but from the earliest days of BB4W it has been apparent that many users find using those libraries far from easy.

I have also, over the years, received a number of comments from those few users of both BBC BASIC and Liberty BASIC (hello Gordon!) to the effect that LB makes it much easier to use GUI features in a program than BB4W does. My initial reaction to such comments (from the perspective of a non-LB user) was to demonstrate that I could produce the same functionality in BB4W. But that missed the point: I could do it, as a very experienced BBC BASIC user, but a lot of people couldn't!

I needed a new perspective on this issue, and that was another reason for taking a more serious interest in Liberty BASIC. Inevitably, in writing the translator from LB to BB4W, I have acquired a far better understanding of LB than I had previously. This has demonstrated two points to me very clearly:

  1. In terms of GUI capability, BB4W is fine: it can do everything that LB can.
  2. In terms of ease of use of GUI support, BB4W (with the existing libraries) is far inferior to LB.
The way I now look at this is in terms of a graph of difficulty versus UI complexity. For the simplest kinds of user interface (the sort of thing one would have done on the BBC Micro) BBC BASIC is extremely easy to use (easier than LB). This isn't surprising given its background in the 1980s: it is richly provided with native text and graphics features. But as one moves towards more complex and 'modern' GUI features (windows, controls, buttons, sprites etc.) the difficulty curve soon overtakes LB. For many features (such as advanced use of sprites, or multiple windows) BB4W effectively lacks any native or library support at all. At this level, resorting to assembler code and direct calls to the Windows API is the only way.

One of the things I would like to trigger a discussion on is: does this matter? Perhaps there is no need to 'compete' with LB. Perhaps it's fine for somebody who wants to use GUI features in his program to migrate to LB rather than using BB4W. On the other hand one might argue that it is letting BB4W users down not to provide GUI libraries which can match the sort of thing it's easy to do in LB.

So what do you think? Do you agree with me that the main weakness of BB4W is not the native language but the libraries? Do you care that it's much easier to implement certain GUI features in Liberty BASIC than in BBC BASIC? What, if anything, should be done about this?

Richard.
Re: The future of BBC BASIC for Windows
Post by David Williams on Feb 29th, 2012, 9:16pm

I'll need a few days to digest and reflect before I can comment, although I might, in the end, prefer not to (comment).

I am definitely interested in reading others' opinions on the points raised in this thread's opening post.

Now, back to finalizing version 2.03 of my little library...


David.
Re: The future of BBC BASIC for Windows
Post by kcpsm on Mar 1st, 2012, 10:16am

Dear Richard,

So a long message in return!

I’m not sure if this will help your thinking or not but I’m delighted to have the opportunity to provide at least some feedback and inputs to you since BB4W has absolutely saved my bacon and made my work so much easier since I bought my first copy in May 2010.

What do people need?

It should go without saying that whatever BB4W (or absolutely anything if you think about it) is, it has to work correctly. Richard, please do not underestimate how much I, and presumably everyone else, values the fact that BB4W does work so well. Hence, I will stand behind you 100% and agree that it is better to maintain the high standard that you have achieved rather than to add new features with the potential to inject bugs (or at least strange effects). With operating systems continuous changing I think you have more than enough on your plate!

Anyway, beyond that, I’m going to narrow this down to just two use models....

a) Compatible with old code.

This is really tricky to pull off completely. When people say they want something to be “the same” then they often don’t really mean that. Running the same code today that used to be run on a 1MHz 8-bit processor with a 40x25 character display can be useful but the chances are it will run way too fast on a PC and the display will be interesting to say the least. Nice form a historical perspective but if the code has further value then it will almost certainly start to be changed to make it suitable for the modern world and machines. So quite how compatible it needs to be is debatable.

I was previously using QuickBASIC for DOS (v4.5). This had the fundamental limit of 64KB of memory so if I have looked for something 100% compatible then it would also have had a memory limit or would have had to interpret the memory manipulation tricks that I needed to use which is ugly as well as pointless.

However, what attracted me to BB4W was that it was ‘compatible with me’. I was (and still am) a SHARP MZ80A owner with Sharps’ own BASIC and Pascal to use in my play and studies. So I’m not from the BBC Micro stream and had never used BBC BASIC before. But what I really saw in BB4W was the true essence of the BASIC language, and just as importantly, that simple working environment where I could just start writing a program and run it. No need to be a computer whiz setting up options and defining windows and memory allocations etc. To cap it all, it just spits out a single EXE file that I can give to anyone and it doesn’t need to be installed with a load of drivers etc.

My field of work is actually hardware design and hardware languages such as VHDL. Whatever anyone claims, there is no such thing as compatible code there either. Even clamped down by rigorous standards of the language itself, for all kind of good and bad reasons people end up with special attributes and directives in their code to make something work with ‘A’ but not with ‘B’. My point being that people say they use a language because it is compatible, portable and timeless etc but in realistic it nearly always needs to be tweaked to fit the final application.

My suggestion – May be if BB4W encountered a keyword from a variant of BASIC that is not supported it could simply present an error message that contained information leading the user to the appropriate equivalent. You already provide so much useful and well written help and information, we just need a pointer to the right place sometimes.


b) To create something absolutely new.

So if someone if going to create something entirely new then compatibility with anything else really shouldn’t be important. Let’s not confuse ‘compatibility’ with being ‘nicely familiar’ so I can make rapid progress.

In my case I did find myself effectively re-writing applications that I had implemented in QuickBASIC in the past but I very quickly got used to any syntax differences. I suppose it would have been nice to just cut and paste in big lumps of old QB code but in reality so much of the code was styled to cope with limitations that were completely gone in BB4W (namely, memory was no longer an issue).

Just about everything I do is based on simple text I/O or communication with a COM port. It always takes a while to learn and convince yourself that you can do these things and once you have a recipe for each the rest is pure fun.

The point I’m trying to make is that once you are writing all new code then you soon get comfortable with the ‘coat you are wearing’ and it becomes the standard against which you measure all others. That is reinforced by the very fact that BB4W works so well and is so easy to ‘wear’. What scares me is what on earth I will do if the next incarnation of Windows comes out and BB4W doesn’t work with it. You see, it is far more important for me to have what we have today in the future rather than adding anything else.

I soon got into BB4W and for the most part made rapid progress because most of my code is similar. Then every now and again I need to do something a bit different and new to me. Call me sad if you like, but this is where the gold old fashioned fun of learning how to do something kicks in. Initially I think something can’t be done but so far there has always been a way with BB4W (even if I had to ask you directly for a clue). There really shouldn’t be a need to add anything to BB4W unless there is absolutely no way of achieving it in a reasonable way. Even then, it should be something that at least more than one person would ever need.

As to ‘Libraries’ then I’m going to admit that I have never used any of them at all! I’m sure there is lots of wonderful stuff in them but I have a personal preference for keeping everything as ‘primitive’ as I possibly can. I guess I’ve just had bad experiences in software and hardware in my past. Anyway, on that basis I won’t comment on libraries as I’m not qualified to do so but I do again want to emphasize how my use model of BB4W is really fundamental and it is that which it services so well.


I hope some of my rambling is of help.

Regards,

Ken
Re: The future of BBC BASIC for Windows
Post by admin on Mar 1st, 2012, 11:02am

on Mar 1st, 2012, 10:16am, kcpsm wrote:
As to ‘Libraries’ then I’m going to admit that I have never used any of them at all! I’m sure there is lots of wonderful stuff in them but I have a personal preference for keeping everything as ‘primitive’ as I possibly can.

I may not be typical, but I almost always want my programs written in BB4W to have the 'look and feel' of a conventional Windows application. To me, the essence of Windows is the standardisation of the User Interface between applications, so that you don't have to learn a new way of selecting files, or a new way of choosing a command from a menu, for every new program you use. I want a user of one of my applications to immediately feel 'at home'. If he happens to prefer a different Windows 'theme', or is disabled and requires 'accessibility' features to be enabled, they should work in the way he is familiar with.

That unavoidably means using the Windows API, since writing your own GUI 'clone' with all the subtleties of keyboard shortcuts, user configuration, accessibility options etc. would be impossible and pointless. Doing it all with SYS calls and assembler code would be well beyond the capabilities of most users, so that's where the libraries come in.

It actually concerns me quite a lot if BB4W encourages people not to use the standard Windows GUI for their programs. I don't want to be reponsible for loads of programs with non-standard User Interfaces coming onto the market!

One of the nice things about Liberty BASIC is that it strongly encourages you to write programs with the Windows look-and-feel. It does provide a crude 'glass teletype' style of interface, but that doesn't support graphics or colours. To do anything at all 'attractive' requires a committment to the Windows GUI. I think that is a very good thing.

So although I can't force anybody to adopt a standardised User Interface, I would like it to be as straightforward as it is in LB so they have no excuse not to, other than being perverse. At the moment that's not the case.

Richard.
Re: The future of BBC BASIC for Windows
Post by Michael Hutton on Mar 3rd, 2012, 7:05pm

"how adequate are the existing libraries supplied with BB4W?"

This is going to be very brief as I have to rush off (not ideal):

1. Keep the old WINLIBX libraries.

2. Make one new large library GUILIB or something similar with all functions needed to make/manipulate windows GUI tools. This may prevent the eternal confusion as to which library I need (WINLIB2 boxes and buttons or is it WINLIB3A or is there a WINLIB4C now.... oh gosh... etc....)

3. Maintain a strict English naming convention for ease of look up for the function names eg

DEF FN_GUILIB_ButtonOnHWND()
DEF FN_GUILIB_ButtonInDialog()
DEF FN_GUILIB_ProgressBarAdvance()
DEF FN_GUILIB_ProgressBarClear()
DEF FN_GUILIB_CreateDialog()
etc etc
or something similar, I personally like Camel Notation. Length of names does not matter.
This make it easier for everyone to contribute functions.

4. Make them all FUNCTIONS so we can see return values.

5. Make a repository where we can all submit functions to add and others can see and edit/correct before they are added to a version.

6. Each function should be documented fully.

7. Only use statics for local variables etc...

6. I have to go but I'll be back!

Michael
Re: The future of BBC BASIC for Windows
Post by admin on Mar 3rd, 2012, 9:29pm

on Mar 3rd, 2012, 7:05pm, Michael Hutton wrote:
Make one new large library GUILIB

I'm attracted to that idea in principle, but it would almost certainly be too big for the trial version of BB4W. We could then end up with a situation where somebody evaluating BB4W to see if it suits his needs isn't able to use the very library that may persuade him of its suitability!

I would be loathe to move to the LB model of having a trial version with unrestricted memory but time-limited. Any thoughts on this issue?

Quote:
Make them all FUNCTIONS so we can see return values

I can't see the point of making something a function, if it inherently doesn't have a value to return. Isn't it more natural for those which do return a value to be functions and those which don't procedures? Looking at LB for inspiration, hardly any of its GUI operations are functions.

Quote:
Each function should be documented fully.

Presumably not within the library itself, otherwise the size issue becomes even more important.

One could conceivably adopt the technique that I have with LBLIB of supplying a crunched version of the library, which is what you actually INSTALL. That's a neat solution, but it imposes quite serious constraints on how the code is written (I get away with it for LBLIB only because the main BASIC program is created automatically, so I know exactly what variables it uses etc.).

Quote:
Only use statics for local variables etc...

Statics? Are you in fact proposing that this library be written in C (i.e. a DLL)? That would certainly solve the 'trial version' issue but it's a drastic step! wink

If it was written in BASIC, as I had anticipated, I would want to avoid PRIVATEs (if that's what you meant) as much as possible because of their adverse interaction with ON ERROR.

Richard.
Re: The future of BBC BASIC for Windows
Post by David Williams on Mar 3rd, 2012, 10:26pm

on Mar 3rd, 2012, 9:29pm, Richard Russell wrote:
I'm attracted to that idea in principle, but it would almost certainly be too big for the trial version of BB4W. We could then end up with a situation where somebody evaluating BB4W to see if it suits his needs isn't able to use the very library that may persuade him of its suitability!

I would be loathe to move to the LB model of having a trial version with unrestricted memory but time-limited. Any thoughts on this issue?


Whilst not myself into GUI programming, I like Michael's idea. What a pity if it doesn't gain wings just because it won't be able to work with the trial version of BB4W due to the limited memory available.

Two ideas:

(1) Write the single, big GUI library, and keep the existing memory-limited BB4W Trial Version model, but write (not single-handedly!) dozens of crystal-clear, well-commented Example Programs showing HOW EASY it is to use BB4W + GUILIB to make GUI-based programs, and publish the programs in HTML form for immediate browsing -- BB4W not required, and for each program, provide a link to a compiled EXE so that the prospective BB4W user can see the program in action. The first dozen-or-so example programs should be very short, and sweet. Then they get progressively longer, complex and more interesting in what they do. These Example Programs, published in HTML form, in a nice visually-pleasing, high-clarity font, should convey ease-of-use and accessibilty in as friendlier way as possible.


Idea #2: Release TWO Trial versions of BB4W, thereby giving prospective customers users the choice between:

A) The usual memory-limited version

B) Significantly more memory (not necessarily as much as 256 MB!), but time-limited. I remember loathing LB's "Nag Screen", and having to wait minutes before being able to use the program, but you don't have to go there. Just do as many other trial-version software packages do: Give the user a total of a few hours (over multiple sessions), and eventually the eggtimer runs out with something like the resulting message: "Thank you for trying BBC BASIC for Windows, but I'm afraid your three-hour Trial Period is now up! If you'd like to purchase..." (or whatever).

Possibly combine Idea #1 with Idea #2, option B: Lots of nice, simple Example Programs in .BBC and HTML form (published and readily browsable on a website), and a time-limited, less memory-constrained, no-nag-screen Trial Version of BB4W.

(And, it goes without saying, that with the alternative Trial Version of BB4W, prospective users can use or try -- ahum -- my library as well).

Michael: Have you changed your e-mail address recently? I tried sending you an e-mail the other day and it bounced!

David.
Re: The future of BBC BASIC for Windows
Post by Michael Hutton on Mar 4th, 2012, 09:48am

"but it would almost certainly be too big for the trial version of BB4W"

I hadn't thought about that. I suppose there is a potential of GUILIB could become rather a monster. I also share your concerns with having to up the amount of memory for the trial version. Hmm. Maybe a GUILIBT version... with minimal but essential functions (or procs!)? And then you could tempt them in with the fact that they get the full version of GUILIB when they buy!
I am not keen on the "supplying a crunched version of the library" for all the restrictions it implies.

"Isn't it more natural for those which do return a value to be functions and those which don't procedures?"

Yes. Agreed. But my thinking was of overall standardisation. I struggle to think of a when a set of API calls don't return a value even if it is just success or failure. It is not a big issue and I would go with what you choose.

Documentation :
"Presumably not within the library itself, otherwise the size issue becomes even more important."

My initial thoughts where at least some documentation in the library. I do tend to ignore size nowadays. It really doesn't matter! And it all gets discarded/crunched anyway. I have just had my fingers burnt so many times for not documenting (or using readable variable names in) a function. I think it would also help in future maintainance. For instance in COMLIB
Code:
      DEF FN_parse(O%,T%,C$,U%)
      LOCAL D%,V%,R%:DIM V% LOCAL 15
      D%=FN_instr(C$,".","(",")",1)
      WHILE D%
        SYS !(!O%+4),O%
        R%=FN_checkargs(O%,3,LEFT$(C$,D%-1),V%)
        IF R% THEN =R%
        SYS !(!O%+8),O%
        O%=V%!8
        C$=MID$(C$,D%+1)
        D%=FN_instr(C$,".","(",")",1)
      ENDWHILE
      =FN_checkargs(O%,T%,C$,U%)

Is surely a nightmare to any future maintainance guy?


"Statics? Are you in fact proposing that this library be written in C (i.e. a DLL)? "

No! LOL, although it could be done.. I probably shouldn't of said it. Ignore that one.


"Release TWO Trial versions of BB4W"
I can only think of the potential nightmares.

David "Have you changed your e-mail address recently?"

Yes. I did. I have sent you a new one. Sorry am slowly getting round to update all my contacts. I hope you understand that my bank was first!

Michael
Re: The future of BBC BASIC for Windows
Post by admin on Mar 4th, 2012, 11:23am

on Mar 3rd, 2012, 10:26pm, David Williams wrote:
These Example Programs, published in HTML form, in a nice visually-pleasing, high-clarity font, should convey ease-of-use and accessibilty in as friendlier way as possible.

I can't help thinking that, however much effort you put into making them attractive etc., they are no real substitute for being able to experiment yourself with programs that actually run in the trial version. After all it's hardly a trial version if it doesn't let you try things!

Quote:
Release TWO Trial versions of BB4W, thereby giving prospective customers users the choice

I must say that really doesn't appeal to me at all. Would you have two different shortcuts on the desktop, or would it initially prompt for which 'limitation' you wanted (memory or time) for that particular session? People would always be trying to 'beat the system' by using the memory-limited option to edit their programs and the time-limited option to test them! Can you imagine constantly switching between the two to optimise your use of the 'free' time? Yuk!

How about attempting a technical solution, such as providing a special 'wrapper' version of GUILIB which contains only the function and procedure entry points (and therefore fits in the trial version's limited memory) and then somehow calls the 'real' library, which is resident in some API-allocated memory block? This would be a bit like the API hack for making large arrays accessible to the trial version, but for PROCs and FNs.

I've not given this any detailed thought, but it's not a million miles away from what CLASSLIB does already to implement Object Classes.

Richard.
Re: The future of BBC BASIC for Windows
Post by admin on Mar 4th, 2012, 3:44pm

on Mar 4th, 2012, 11:23am, Richard Russell wrote:
I've not given this any detailed thought, but it's not a million miles away from what CLASSLIB does already to implement Object Classes.

OK, here's a proof of concept. The GDI+ library GDIPLIB is too large to load in the trial version, with the default setting of HIMEM. I have put a wrapper library GDIPLIB_ (using a simple adaptation of the code in CLASSLIB) here; download it to your @lib$ folder:

http://tech.groups.yahoo.com/group/bb4w/files/Libraries/GDIPLIB_.BBC

You should now find that the GDI+ demo program below runs successfully in the trial version:

Code:
      REM. Demonstration of GDI+ library
      
      LineEndSquare = 1
      LineEndRound = 2
      LineEndTriangle = 3
      LineEndSquareAnchor = &11
      LineEndRoundAnchor = &12
      LineEndDiamond = &13
      LineEndArrow = &14
      
      LineStartSquare = &100
      LineStartRound = &200
      LineStartTriangle = &300
      LineStartSquareAnchor = &1100
      LineStartRoundAnchor = &1200
      LineStartDiamond = &1300
      LineStartArrow = &1400
      
      LineDash = &10000
      LineDot  = &20000
      LineDashDot = &30000
      LineDashDotDot = &40000
      
      INSTALL @lib$+"GDIPLIB_"
      
      VDU 23,22,640;480;8,16,16,128
      
      PROC_gdipinit
      
      REM Near-horizontal line to demonstrate antialiasing
      pencolour% = &FFFF0000 : REM. Opaque red
      penstyle% = LineStartArrow + LineEndArrow
      penwidth = 4.0
      pen% = FN_gdipcreatepen(pencolour%, penstyle%, penwidth)
      PROC_gdipline(pen%, 0, 200, 1280, 210)
      PROC_gdipdeletepen(pen%)
      
      REM Near vertical line to demonstrate antialiasing
      pencolour% = &FF008000 : REM. Opaque dark green
      penstyle% = LineDashDot
      penwidth = 2.0
      pen% = FN_gdipcreatepen(pencolour%, penstyle%, penwidth)
      PROC_gdipline(pen%, 200, 0, 210, 960)
      PROC_gdipdeletepen(pen%)
      
      REM Bezier curve from four control points
      pencolour% = &FF0000FF : REM. Opaque blue
      penstyle% = LineStartRound + LineEndArrow
      penwidth = 5.0
      pen% = FN_gdipcreatepen(pencolour%, penstyle%, penwidth)
      PROC_gdipbezier(pen%, 20, 20, 200, 200, 400, 600, 800, 600)
      PROC_gdipdeletepen(pen%)
      
      REM Ellipse made from 360-degree arc
      pencolour% = &A0FF8000 : REM. Translucent orange
      penstyle% = 0
      penwidth = 3.0
      pen% = FN_gdipcreatepen(pencolour%, penstyle%, penwidth)
      PROC_gdiparc(pen%, 500, 500, 150, 100, 0, 360)
      LINE 500, 500, 500, 500
      PROC_gdipdeletepen(pen%)
      
      REM Filled sector, 3/4 of an ellipse
      brushcolour% = &FFFF00FF : REM. Opaque magenta
      brush% = FN_gdipcreatebrush(brushcolour%)
      PROC_gdipsector(brush%, 900, 800, 100, 150, 90, 270)
      LINE 900, 800, 900, 800
      PROC_gdipdeletebrush(brush%)
      
      REM Filled five-pointed star
      brushcolour% = &FF808080 : REM. Opaque grey
      brush% = FN_gdipcreatebrush(brushcolour%)
      DIM X(4), Y(4)
      X() = 1100,900,1060,1000,940
      Y() = 430,430,320,500,320
      PROC_gdippolygon(brush%, 5, X(), Y(), 0)
      *GSAVE alternate 900,320,200,180
      PROC_gdippolygon(brush%, 5, X(), Y(), 1)
      *GSAVE winding 900,320,200,180
      PROC_gdipdeletebrush(brush%)
      
      REM Square drawn with polyline
      pencolour% = &FF000000 : REM. Opaque black
      penstyle% = 0
      penwidth = 2.0
      pen% = FN_gdipcreatepen(pencolour%, penstyle%, penwidth)
      DIM xx(4), yy(4)
      xx() = 500,500,600,600,500
      yy() = 700,800,800,700,700
      PROC_gdippolyline(pen%, 5, xx(), yy())
      PROC_gdipdeletepen(pen%)
      
      REM Angled ellipse drawn with polybezier
      pencolour% = &FF00FF00 : REM. Opaque bright greenn
      penstyle% = 0
      penwidth = 2.0
      pen% = FN_gdipcreatepen(pencolour%, penstyle%, penwidth)
      PROCgdipellipse(pen%, 350, 800, 50, 100, PI/4)
      PROC_gdipdeletepen(pen%)
      
      PROC_gdipexit
      END
      
      DEF PROCgdipellipse(pen%, x, y, a, b, t)
      LOCAL c, d, o, p, q, r, x(), y()
      DIM x(12), y(12)
      c=a*SIN(t) : a=a*COS(t)
      d=b*SIN(t) : b=b*COS(t)
      o=a*0.552285 : q=c*0.552285
      p=b*0.552285 : r=d*0.552285
      x() = x-a,x-a-r,x-o-d,x-d,x+o-d,x+a-r,x+a,x+a+r,x+o+d,x+d,x-o+d,x-a+r,x-a
      y() = y+c,y-p+c,y-b+q,y-b,y-b-q,y-p-c,y-c,y+p-c,y+b-q,y+b,y+b+q,y+p+c,y+c
      PROC_gdippolybezier(pen%, 13, x(), y())
      ENDPROC

Obviously this technique could be abused to work around the limitations of the trial version, but as with the array hack if somebody is prepared to go to that amount of trouble they probably deserve to succeed!

If you haven't already got both full and trial versions installed (as 'power' users I would think it likely you have!) see the instructions here:

http://bb4w.wikispaces.com/Installing+both+demo+and+full+versions

Richard.
Re: The future of BBC BASIC for Windows
Post by Michael Hutton on Mar 4th, 2012, 10:57pm

Wow! Very good!

I notice you used a fail safe so that it can be used with every library. In the case of GDILIB you need to call the init first so in reality the GDILIB_ Wrapper could contain only:
Code:
      REM Wrapper for GDIPLIB
      
      DEF FN_gdipcreatepen(C%, S%, w)
      DEF FN_gdipcreatebrush(C%)
      DEF PROC_gdipdeletepen(P%)
      DEF PROC_gdipdeletebrush(B%)
      DEF PROC_gdipline(P%, x1, y1, x2, y2)
      DEF PROC_gdippolyline(P%, N%, x(), y())
      DEF PROC_gdipbezier(P%, x1, y1, x2, y2, x3, y3, x4, y4)
      DEF PROC_gdippolybezier(P%, N%, x(), y())
      DEF PROC_gdiparc(P%, xc, yc, xr, yr, as, ad)
      DEF PROC_gdipsector(B%, xc, yc, xr, yr, as, ad)
      DEF PROC_gdippolygon(B%, N%, x(), y(), M%)
      DEF FN_gdipg
      DEF PROC_gdipinit : PROC_gdip_redirect : PROC_gdipinit : ENDPROC
      DEF PROC_gdipexit
      DEF PROC_bbc2api(RETURN x, RETURN y)
      DEF FN_f4(A#)

etc...

or
Code:
      REM Wrapper for GDIPLIB
      
      DEF FN_gdipcreatepen(C%, S%, w)
      DEF FN_gdipcreatebrush(C%)
      DEF PROC_gdipdeletepen(P%)
      DEF PROC_gdipdeletebrush(B%)
      DEF PROC_gdipline(P%, x1, y1, x2, y2)
      DEF PROC_gdippolyline(P%, N%, x(), y())
      DEF PROC_gdipbezier(P%, x1, y1, x2, y2, x3, y3, x4, y4)
      DEF PROC_gdippolybezier(P%, N%, x(), y())
      DEF PROC_gdiparc(P%, xc, yc, xr, yr, as, ad)
      DEF PROC_gdipsector(B%, xc, yc, xr, yr, as, ad)
      DEF PROC_gdippolygon(B%, N%, x(), y(), M%)
      DEF FN_gdipg
      DEF PROC_gdipinit : PROC_gdip_redirect : PROC_gdipinit
      DEF PROC_gdipexit
      DEF PROC_bbc2api(RETURN x, RETURN y)
      DEF FN_f4(A#)
      ENDPROC

Although I understand that many libraries do not have an init PROC and so you would have to ensure that you do what you did.

All in all though, great. Definitely the way to make a large library available without having to change anything in the trial version.

I particularly like the !EVAL() which is also so essential in CLASSLIB.

Quote:
Each function should be documented fully.

Now the full library could be as verbose as you like.

Michael
Re: The future of BBC BASIC for Windows
Post by admin on Mar 9th, 2012, 1:38pm

So, from the general lack of response, I suppose I must conclude that hardly anybody else shares my concern that BBC BASIC is significantly inferior to Liberty BASIC in some important respects. This is disappointing, but not altogether unexpected.

On that basis I have to assume there is little enthusiasm for a new GUILIB library. It is my opinion that without such a library BBC BASIC should not be promoted as a 'Windows' programming language, since it does not adequately support the GUI paradigm that is the essence of Windows.

Richard.

Re: The future of BBC BASIC for Windows
Post by Michael Hutton on Mar 9th, 2012, 5:28pm

on Mar 9th, 2012, 1:38pm, Richard Russell wrote:
.... shares my concern that BBC BASIC is significantly inferior to Liberty BASIC in some important respects.


I am sorry, I am just not able to comment not being a LB user.

I am willing to help get a GUILIB up and running. But with this I would not be able to compare it to LB at all.

I don't know who else in the world of BB4W also uses LB except for Gordon and you, and the few posters in the yahoo groups.

Michael
Re: The future of BBC BASIC for Windows
Post by admin on Mar 9th, 2012, 10:01pm

on Mar 9th, 2012, 5:28pm, Michael Hutton wrote:
I am sorry, I am just not able to comment not being a LB user.

Although one could use the LB GUI model as a template for the sort of features a new GUILIB should provide, an equally valid template might be the Visual Basic model, or any other GUI framework. I would welcome suggestions from all sources.

What is clear is that the existing WINLIBx libraries aren't satisfactory. They were some of the earliest BB4W libraries ever written, at a time when my understanding of the Windows GUI was poor and my experience with other GUI frameworks was zero! If BB4W is to be competitive it needs an entirely new approach to supporting the Windows GUI.

One thing that worries me, though, is that several people in this group (and the wider BB4W user community) probably have no experience of any other GUI framework. They therefore cannot see just how awful the existing WINLIBx libraries are, because they are all they have ever known! This is a recipe for disaster.

Perhaps I could ask everybody reading this (yes, that means you) to give a brief resume of their experiences with other Windows GUI frameworks and what they can bring to the party if we are to get this GUILIB idea off the ground.

Richard.
Re: The future of BBC BASIC for Windows
Post by Andy Parkes on Mar 12th, 2012, 11:37am

Hi All,

As I was reading through the messages thus far, I began by thinking that the WINLIBx libraries were adequate for their respective purposes. But even so, over the years I have written several simple functions and procedures to facilitate my use of these libraries, typically things that just take the monotony out of enabling dialogue controls etc.

I am sorry to say that I have no experience of other Windows GUI frameworks, and I had assumed that WINLIBx was about as easy as it got. However, I have long since noticed the limitations of WINLIBx. As Richard said, at the beginning, difficulty increases a lot, with user interface complexity. You guys - and some other folks - have written very helpful articles in the BB4W Wiki. Although I have been able to make good use of this, and increase my repertoire, there is a lot of useful work in both in the Wiki and the reference manual, which could be put at people’s fingertips by structuring it in a library. I think a 'List View' control is but one example of something crying out for library support.

I agree with Richard’s suggestion that the limitations of the library interface to the GUI, is one of the weaker aspects of BB4W. GUI programming does matter. I think most programmers get their kicks out of seeing what other people can do with their programs. Most people use Windows, most people want the Windows GUI.

I take Richards point that I (we) may be slightly delusional about the relative value of WINLIBx. Let’s first take GUILIB as far as we can, then see how we feel about WINLIBx.

The trial version of BB4W without the time limit is most helpful, and makes it easy for someone to use BB4W as a platform for teaching programming.

It is difficult for me to say at this point just how I might be of service. I'm certainly not going to be pushing the frontiers of this, but I can probably reconstitute things that are already within the scope of the Wiki and manual, into the given library format. I’m still (slooowly!) learning assembly language, so I won’t be doing anything technical!

If I was doing something like this myself, I would start with a list of what I would like to include (with descriptions for each item, stating how I would like to see it implemented – that is to say, primarily a description of how it should look from the point of view of someone using the library).

This idea is rather jumping the gun, but if it is the case that GUILIB ends ups comprising hundreds of functions and procedures, it might be nice to have a tool, which weeds out the hundred or so which are not actually used by a particular program. A good example of this is DrWIMP , another BASIC library that is similar in concept to that which we are now proposing. DrWIMP is a 'GUILIB' for use with RISC OS BASIC V. RISC OS does not offer the variety that Windows does, yet even this single library contains approximately 292 functions and procedures. It gets past the drag factor with a linker.

I have not given this much thought yet, but DrWIMP is perhaps a good example of a library that makes event driven programming feel ‘natural’ in an intrinsically procedural programming language. The library is supplied with a skeleton application. The skeleton application contains predefined functions, which are called by the library depending upon the event which needs handling. I can’t remember the correct names of the functions, but as you would expect there is something along the lines of, FNuser_mouseClicked, FNuser_mouseRollOver, FNuser_mouseEnterWindow, FNuser_mouseLeaveWindow, FNuser_windowMoved, FNuser_windowResized etc. Then all the programmer has to do is fill each function with code to filter the window handle, button handle, mouse button used (all passed to each function) etc., and generate the appropriate response. The values returned from these functions provide a way of optionally returning values to the DrWIMP library depending upon the outcome of the filtered mouse click etc.

It’s been a while since I last used it, but I do recall DrWIMP being very easy to use, and also that it was a significant enabler under RISC OS, partly because BASIC V does not have something equivalent to ON SYS etc., also the RISC OS API seems to me now, to be more difficult than the Windows API, but that might just be the effect of time.

It’s probably worth mentioning that I'm ISP challenged (connection today is courtesy of my very noisy local library). I can't say when I'll be back, but I will be! In the mean time, I'm going to download and install the trial version of LBB to see if there is anything I can glean.

Regards

Andy Parkes
Re: The future of BBC BASIC for Windows
Post by DDRM on Mar 13th, 2012, 09:54am

Hi Richard,

Sorry for the delay in responding, but it’s been rather a busy time. I AM interested, and I DO think there is potential for improvement.

One comment before I start on the main response. I suspect you tend to see BB4W as a way of producing high-class applications for the Windows environment, and so it is. Making that easier is highly desirable, and will undoubtedly increase its appeal. However, I suspect that many, perhaps most, users are beginners or relatively low level hobbyists, for whom the simplicity and immediacy of the language is very appealing. The ability to give one or two parameters, do some calculations, and then plot a simple graph, all in the main window, and all in a handful of lines of code, is very rewarding. I would be very sad if we lost that simple, basic, underlying approach in moving more towards a GUI-oriented approach, where one has to faff about getting the interface sorted before you can do anything with it!

In response to your original question on this strand, I agree that BB4W can do virtually everything I would want to do in the context of Windows programming, but that the library support for API commands could be better, and make life easier. Even more detailed Help would help - for example, the Help for many of the libraries gives examples of the style codes to do obviously helpful things, but not the full list. I appreciate adding ALL options in each place isn’t practical, but it does mean often going away and having to find (somewhere else) what those options, and either the values or the variable names, are.

Most of my programming experience is in various dialects of BASIC, going back into the 80s, though I was also competent (in my own, rather limited terms) in Turbo Pascal. In terms of languages that used the Windows interface, I have really only used Visual Basic and BB4W. Probably because of my background, I HATED VB, because it was SO focussed on event-driven programming, and a lot of the time I was interested in programs which would be doing a lot in the background, and then showing me the results - for me, then, finding BB4W was a blessing and a great relief. Having said that, the VB way of doing things - design the interface elements, and then put code in to make them all work - works very well for that kind of application.

One thing that I particularly liked in VB was that one could set or read parameters of a control directly - so, for example (and forgive me if the syntax is not exactly correct - I don’t have VB on this computer, and have barely used it since discovering BB4W) to set or read the contents of an edit control on simply wrote

editbox1.text=”Hello world!”

Indeed, one can use them as variables in an expression. This is much more user friendly (to my mind) than having to use

SYS "SetDlgItemText", !dlg%, id%, text$.

My guess is that this would be difficult to implement directly in BB4W, and is what you mean when you say that LB is different because it makes the GUI components an intrinsic part of the language (or something like that - can’t find the reference to hand). I wonder though whether it would be possible to handle this via a level of abstraction. I haven’t thought through how this would have to work, but could you have a structure associated with each control, containing its key (user settable/readable) characteristics, and then hide all the systems calls in the background, via a library? That might mean

PROCGUI_Write(editbox1.text,”Hello world!”) , or boxcontents$=FNGUI_Read(editbox1.text$),

for example. Then each time you create a new control, an associated structure would be made, containing default parameters (perhaps filled via DlgEdit or an equivalent).

Would it be possible and useful to provide a similar abstraction for MODE, PLOT and some of the VDU codes? For example, to be able to
PROCGraphics_SetMainWindow(xres%,yres%,bigchars%,colours%)
PROCGraphics_Trianglefill(x1%,y1%,x2%,y2%,x3%,y3%,colour%).

Best wishes,

David
Re: The future of BBC BASIC for Windows
Post by admin on Mar 13th, 2012, 11:51am

on Mar 13th, 2012, 09:54am, DDRM wrote:
The ability to give one or two parameters, do some calculations, and then plot a simple graph, all in the main window, and all in a handful of lines of code, is very rewarding. I would be very sad if we lost that simple, basic, underlying approach

I can't see that there's any prospect of 'losing' that ability (after all I'm not proposing any changes to BB4W itself!). As the likes of Jonathan Harston emphasise, graphics are not actually 'part' of BBC BASIC anyway: they are an OS feature which is emulated in BB4W to improve compatibility with the BBC Micro and Acorn Archimedes.

Richard.
Re: The future of BBC BASIC for Windows
Post by admin on Mar 13th, 2012, 12:02pm

on Mar 13th, 2012, 09:54am, DDRM wrote:
Then each time you create a new control, an associated structure would be made

You seem to be talking about Object Oriented programming here. I have wondered to what extent a possible GUILIB could/should be integrated with CLASSLIB, but I can see potential for an adverse reaction from users if they are forced to learn and use the (somewhat kludgey) OO syntax that CLASSLIB requires. Bear in mind that CLASSLIB (as recently improved by Jon Ripley) represents the 'state of the art' as far as a BB4W OO library is concerned; I don't think it's possible to do much better without modifying the language.

Richard.
Re: The future of BBC BASIC for Windows
Post by admin on Mar 13th, 2012, 5:46pm

on Mar 13th, 2012, 09:54am, DDRM wrote:
to set or read the contents of an edit control on simply wrote
Code:
editbox1.text=”Hello world!”

That's similar to the Liberty BASIC syntax:
Code:
#window.editbox1 ”Hello world!”

but of course BBC BASIC has nothing similar, and modifications to the language are not on the agenda.

Options one would have in BBC BASIC include:

The current way:
Code:
SYS "SetWindowText", hctrl%, text$

A minimal thin wrapper (basically just a cheat to make it look more native):
Code:
PROC_SetText(hctrl%, text$)

A more hierarchical approach:
Code:
PROC_SetText(hdlg%, idctrl%, text$)

Using a structure-like syntax (but I'm not sure this is practical, because BB4W doesn't support creating a structure dynamically):
Code:
PROC_SetText(dlg.ctrl, text$)

The Object Oriented approach e.g. using CLASSLIB:
Code:
PROC(ctrl.SetText)(text$)


Do any of these particularly appeal? If not, have you any suggestions for an alternative?

Richard.
Re: The future of BBC BASIC for Windows
Post by DDRM on Mar 14th, 2012, 10:05am

Hi Richard,

I guess it could be a bit like OOP, but that wasn't really what I had in mind (I don't want to encapsulate the methods with the data as an object) - I envisioned a set of functions and procedures with a "BB4W feel" to them, more like your second option:

Code:
PROC_SetText(hctrl%, text$)


In this particular case, as you say, it is largely cosmetic, but PROC_ReadText will be a bit more complicated (though still given in the manual), especially for multi-line dialogue boxes, and other things are more complicated still.

In VB you can change the size/position/characteristics of windows and controls directly, in a similar way, and it would be nice to be able to do this in BB4W, too.

My “level of abstraction” was the idea that you could have ?an array of structures? (size set when you call GUILibInit) in the background that stored the details for each control when it was created. For example, it would store which window/dialogue box handle it was associated with, so you don’t need to ask for hdlg%, and what control type it is, so that the routines can handle them appropriately. Of course it would also need to store content, any constraints (bounds for scroll bars etc), position, size, style, etc.

The idea would be that to access/change information relating to a control, you would use the appropriate function or procedure, with either the id% or a more user-friendly alias: “client_name”; again the alias would be stored in the structure and looked up when required. Alternatively, and faster I presume, someone cleverer than me could write a utility [pedant mode] should that be AN utility? [/pedant mode] to convert all the aliases to their equivalent IDs, a la the Windows constants utility.

Using an alias might actually require the parameter to be a quoted string, which is a bit ugly: it would be nicer if you could treat it like a variable name, but I’m not sure I could disentangle that without causing a “no such variable” error. Maybe you could...

Would it be possible/useful/desirable to cause controls which need initialising AFTER a dialogue box is shown to have that happen automatically with PROC_showdialogue? The relevant information could be provided when the control is first created, and then PROC_showdialogue could look up the associated controls, and initiate them as required?

As you have said, all these things CAN be done in BB4W, but a lot of people seem to be scared of even CONSIDERING using SYS calls, even when it is a straight cut and paste from the manual. It’s largely about wrapping these complexities up in a familiar, consistent form.

Hope that makes some sort of sense....

D
Re: The future of BBC BASIC for Windows
Post by admin on Mar 14th, 2012, 2:49pm

on Mar 14th, 2012, 10:05am, DDRM wrote:
I guess it could be a bit like OOP

VB is Object Oriented.

Quote:
In VB you can change the size/position/characteristics of windows and controls directly, in a similar way, and it would be nice to be able to do this in BB4W, too.

I have never programmed in VB, so can you give an example of the sort of syntax you have in mind? Changing the size/position/style of a control after a dialogue box has been opened is not something I had anticipated supporting (other than via direct API calls).

Quote:
For example, it would store which window/dialogue box handle it was associated with, so you don’t need to ask for hdlg%, and what control type it is, so that the routines can handle them appropriately. Of course it would also need to store content, any constraints (bounds for scroll bars etc), position, size, style, etc.

Are you proposing dispensing with the conventional dialogue box template and creating the dialogue box in an alternative, non-standard, way? Admittedly that's exactly what LBB does, but I hadn't envisaged GUILIB departing so radically from the 'norm'.

Quote:
Would it be possible/useful/desirable to cause controls which need initialising AFTER a dialogue box is shown to have that happen automatically with PROC_showdialogue?

What would be the advantage?

Quote:
The idea would be that to access/change information relating to a control, you would use the appropriate function or procedure, with either the id% or a more user-friendly alias: “client_name”

Because it will always be necessary to do some things using direct Windows API (SYS) calls - that's inevitably the case in both VB and LB too - it will be important to ensure that the design of any GUILIB library doesn't make that more difficult than it needs to be. That suggests to me that it will always be better to refer to controls in ways that Windows understands - i.e. handles and/or IDs - rather than some BB4W-specific alias notation.

Quote:
Hope that makes some sort of sense....

Because of my lack of knowledge of Visual Basic many of the concepts you have described are unfamiliar to me. Can you perhaps illustrate them with some examples in a suggested BB4W syntax?

Richard.
Re: The future of BBC BASIC for Windows
Post by DDRM on Mar 15th, 2012, 3:33pm

Hi Richard,

Quote:
I have never programmed in VB, so can you give an example of the sort of syntax you have in mind?


here’s a clip in VB:
Code:
Form4.Left = Form2.Left
Form4.Top = Form2.Top + Form2.Height
Form4.Visible = True
drawgraphs
calclanes
sumlanes
inited = True
Form3.Left = Form2.Left + Form2.Width
Form3.Top = Form2.Top
Form3.Visible = True

Here I have read a picture of (unknown) size into form2 and displayed it, and I can then position form4 beneath it, and form 3 to the right of it. Form4 will contain some densitometry data from the picture (of an electophoresis gel), while form3 is the controls for subsequent processing (adjusting lane positions, etc).

Quote:
Changing the size/position/style of a control after a dialogue box has been opened is not something I had anticipated supporting (other than via direct API calls).


In another program, I changed the size of the form dynamically to hide or reveal some of the controls - effectively an "easy" or "advanced" interface. By changing the control.visible parameter you can also hide or reveal controls. I think this is really convenient. I guess it is pretty simple to hide or show components, as it is windows, so it would be a pretty thin wrapper!

Quote:
Are you proposing dispensing with the conventional dialogue box template and creating the dialogue box in an alternative, non-standard, way? Admittedly that's exactly what LBB does, but I hadn't envisaged GUILIB departing so radically from the 'norm'.


No, I thought creating the dialogue box and the controls would be essentially unchanged: the only difference would be that the functions creating them would stick the data into an array of structures, so that subsequent calls to associated functions could look up the data they needed.

Quote:
What would be the advantage?

[of automatically initialising controls when a dialogue box is made visible]. Well, you could set everything up in one place at one time.

Quote:
Because it will always be necessary to do some things using direct Windows API (SYS) calls - that's inevitably the case in both VB and LB too - it will be important to ensure that the design of any GUILIB library doesn't make that more difficult than it needs to be. That suggests to me that it will always be better to refer to controls in ways that Windows understands - i.e. handles and/or IDs - rather than some BB4W-specific alias notation.

I think that’s probably right, and of course all the windows controls would still have handles/IDs. I just think it’s easier to remember that a control is called “client_name” rather than that its ID is 103. It wouldn’t be either/or, and indeed you would probably want to have equivalent functions:

name$=FN_GUI_GetTextN(“client_name”)
name$=FN_GUI_GetText(103)

Quote:
Because of my lack of knowledge of Visual Basic many of the concepts you have described are unfamiliar to me. Can you perhaps illustrate them with some examples in a suggested BB4W syntax?


I’ll try and do a small example, but I can’t promise it today (I have a large chunk of a draft project to read for a student I’m seeing tomorrow morning...).

Best wishes,

D
Re: The future of BBC BASIC for Windows
Post by admin on Mar 15th, 2012, 6:05pm

on Mar 15th, 2012, 3:33pm, DDRM wrote:
here’s a clip in VB

I was hoping for some proposed BBC BASIC syntax. Only in an OO language will a simple 'assignment' like:

Code:
Form4.Left = Form2.Left

have an 'action'. In BBC BASIC it fundamentally cannot do anything other than to set the variable 'Form4.Left' to a new value.

Quote:
I guess it is pretty simple to hide or show components, as it is windows, so it would be a pretty thin wrapper!

I cannot see the point of thin wrappers which add no functionality; to me they are just a means of obfuscating the Windows API. If the user learns how to do something by calling the Windows API directly it's a valuable skill which will serve him well if he advances to using another language, like C++. On the other hand learning a BB4W-specific syntax is a dead-end; if he moves to another language he will have to forget it and start again from scratch.

Quote:
No, I thought creating the dialogue box and the controls would be essentially unchanged

That implies quite a lot of data duplication (parameters of controls would have to go into the dialogue box template and into your 'structures'). I'm always uneasy about duplication of this sort; it's inelegant and you risk obscure side-effects if the two copies get out-of-step.

I think this thread has come to the end of the road. My thanks to everybody who has taken the trouble to contribute.

Richard.
Re: The future of BBC BASIC for Windows
Post by DDRM on Mar 28th, 2012, 10:48am

Hi Richard,

I know you have declared this strand closed, but I have continued to have a bit of a play at the kinds of things I was interested in.

I started by looking at modifying the dialogue box template directly, as you implied, starting with Winlib2 as a teaching aid, but I have strayed away from that idea, because what I want is to be able to manipulate the dialogue box on the fly - I often use a dialogue box as a controls window, and leave it open all the time, but I may wish to be able to provide a simpler or more complex interface, depending on context.

I agree with you to some extent about providing very thin wrappers to obscure the API from the user, but I thought that the discussion started off by acknowledging that one could do virtually anything you wanted to in BB4W, but that many people were scared to do so, and we wanted to make it more approachable. A further counter-argument is that by providing them with a library (the code of which they can inspect) they can use it easily if that's all they want, or they can use it as a teaching aid, to see (a) how easy it is, and (b) have a sample of what to do. For example, the scaling of locations in dialogue box units is straightforward once you understand how to do it, but it took me quite a long time (even after I'd found Michael Hutton's posts on the Yahoo forum, which suggest he had also banged his head on it for a while). Another thing I found tricky was finding the handle for the controls: once I knew how to do it, I found that it is in fact in the manual, but it wasn't obvious to me.

I have posted a file called "potential winlib2 extensions.bbc" in the libraries section of the files section of the yahoo forum. This is essentially a testbed, but includes a set of small procedures and functions that I think might be more generally useful.

The program in itself doesn't do anything useful (draws some shapes in the main window, purely as an indication that the buttons etc are doing what they are meant to. Controls jump around in the dialogue box in an alarming and very annoying way - of course one would never actually do this, but it is just to test out the functions.

In some cases I have provided a proc/function as a very thin wrapper (for example, PROCsetdlgtext) - in that case because it is a logical pair to FNgetdlgtext (which is stolen straight from the manual). In others, such as GetWindowRect, I haven't bothered, which may be inconsistent - one could of course easily add these.

I'd be interested to know if anyone else thinks this kind of thing would be useful.

Best wishes,

David