Is FreeBasic still continued to develop?

[Landeel]

This is great news!!! Thanks v1ctor!!! You're the best!!!

[John Spikowski]

Welcome FreeBASIC to the Basic to C club.

Current Members:

BCX
BaCon

[v1ctor]

Just to make it clear, FB won't just translate BASIC to C. Gcc is used as a high-level assembler, not a single C header is ever included - you can also use OOP that's ABI-compatible with g++ (a C++ compiler) using gcc to compile (a C-only compiler).

The gcc emitter doesn't change anything in the front- and the middle-end of the compiler, so syntax and semantic checks continue to be done.

[John Spikowski]

not a single C header is ever included

Can C headers be included to add other libraries like Gtk, cURL, ...?

[v1ctor]

No, inline C isn't and won't be supported.

Most C libraries can already be used in FB, including those listed. To support syntax and semantic checks and debugging (what a simple translator won't do), the headers must be in the FB language. The tool: SWIG.

[Dr_D]

long live fb!

[joseywales72]

Great news.... Thank you Victor and all the devs.

[HD_]

I love you v1ctor :D

[fsw]

Just to make it clear, FB won't just translate BASIC to C. Gcc is used as a high-level assembler, not a single C header is ever included - you can also use OOP that's ABI-compatible with g++ (a C++ compiler) using gcc to compile (a C-only compiler).

The gcc emitter doesn't change anything in the front- and the middle-end of the compiler, so syntax and semantic checks continue to be done.

This is so cool!

[SARG]

@v1ctor

I have some questions concerning the gcc emitter.

I'm not sure to understand "Gcc is used as a high-level assembler". Could you explain a bit more ?

About the stabs informations what changes ? Are they always stored in the same way ?

Currently is it possible to do tests of debugging ? And how ?

Thanks for the done job and for your answers.

[v1ctor]

Like this:

Code: Select all

	var a = 1, b = 2
	print a + b

Becomes:

Code: Select all

	A$ = (integer)1;
	B$ = (integer)2;
	#define vr$1 ((integer)(A$ + B$))
	fb_PrintInt( (integer)0, vr$1, (integer)1 );

About stabs, the current implementation is based on the debugging info that gcc 3.x generates, so i think that they are compatible. I've no idea if they added anything new to gcc 4.x.

Yeap, but i wouldn't do any big tests atm. I committed the fixes to allow debugging in -gen gcc today, they are in the gengcc branch.

[D.J.Peters]

hi v1ctor nice to see you have time for your baby FB :-)

FB versus GCC :lol:

Code: Select all

' The Lord of the Julia Rings
' The Fellowship of the Julia Ring
' Free Basic
' Relsoft
' Rel.BetterWebber.com

'#define USE_PTC

#ifdef USE_PTC
  #define PTC_WIN
  #include once "tinyptc.bi"
#endif

const SCR_WIDTH  = 320
const SCR_HEIGHT = 240
const SCR_SIZE = SCR_WIDTH*SCR_HEIGHT
const SCR_MIDX = SCR_WIDTH  \ 2
const SCR_MIDY = SCR_HEIGHT \ 2


const as single PI = 3.141593
const MAXITER = 20

dim shared as single Lx(SCR_WIDTH -1)
dim shared as single Ly(SCR_HEIGHT-1)
dim as uinteger ptr pBuffer
dim shared as uinteger Buffer(SCR_SIZE-1)

#ifdef USE_PTC
  if ptc_open( "Julia (Relsoft) TinyPTC", SCR_WIDTH, SCR_HEIGHT ) = 0 then
    print "error: ptc_open() !"
    beep:sleep:end
  end if
  pBuffer = @Buffer(0)

#else
  screenres SCR_WIDTH,SCR_HEIGHT,32
  if screenptr = 0 then
    print "error: screenres !"
    beep:sleep:end
  end if
  WindowTitle "Julia (Relsoft) FBGfx"
  pBuffer = ScreenPtr()
#endif

dim as uinteger ptr pTop,pBottom
dim as single  xmin, xmax,ymin,ymax
dim as single  theta,ty,p,q
dim as single  x,y,x2,y2
dim as integer px,py,i,i_last,ii
dim as integer frames
dim as integer red,grn,blu
dim as double  sTime,nTime
dim as single  cmag,cmag2,zmag,zTot
dim as single  drad,drad_L,drad_H

xmin = -2.0:xmax =  2.0
ymin = -1.5:ymax =  1.5

for i = 0 to SCR_WIDTH - 1
  lx(i) = xmin + i * (xmax - xmin) / (SCR_WIDTH  - 1)
next i

for i = 0 to SCR_HEIGHT - 1
  ly(i) = ymax - i * (ymax - ymin) / (SCR_HEIGHT - 1)
next i

sleep 1000,1

' start time
sTime = Timer()
do
  pTop    = pBuffer
  pBottom = pBuffer + (SCR_SIZE-1)
  frames = (frames + 1) and &H7fffffff
  theta = frames * PI / 180
  p = cos(theta) * sin(theta * .7)
  q = sin(theta) + sin(theta     )
  p = p * .6
  q = q * .6
  cmag2 = (p*p + q*q)
  cmag = sqr(cmag2)
  
  drad = 0.04
  drad_L = cmag   - drad
  drad_L = drad_L * drad_L
  drad_H = cmag   + drad
  drad_H = drad_H * drad_H

#ifndef USE_PTC
  ScreenLock
#endif

  for py = 0 to SCR_HEIGHT\2 -1
    ty = ly(py)
    for px = 0 to SCR_WIDTH - 1
      x = Lx(px)
      y = ty
      ztot =0
      i = 0
      while (i < MAXITER)
        x2 = x * x
        y2 = y * y
        zmag = (x2 + y2)
        if zmag > 4.0 then exit while
        if (i > 0) then
          if (zmag > drad_L) then
            if (zmag < drad_H) then
              ztot = ztot + ( 1 - (abs(zmag - cmag2) / drad))
              i_last = i
            end if
          end if
        end if
        i = i + 1
        y = x * y
        y = y + y + q
        x = x2-y2 + p
      wend
      
      if ztot > 0 then
        i = cint(sqr(ztot) * 500)
        if i>0 then
          if i<256 then
            red=i:  i=i * 0.33
            red=(red+i) * 0.33
            grn=     i  * 0.33
            blu=(red+i) * 0.33
          else
            i-=256
            if i<256 then
              grn=i:i=(255+i) * 0.33
              red=(255+grn+i) * 0.33
              grn=(grn+    i) * 0.33
              blu=(red+    i) * 0.33
            else 
              i-=256
              if i<256 then
                blu=i:ii=(510+i) * 0.33
                red=(510+    ii) * 0.33
                grn=(255+i  +ii) * 0.33
                blu=(i  +red+ii) * 0.33
              else
                red=254 '= (255+255+(255+255+255)*0.33)*0.33
                grn=254 '=      "                 "
                blu=254 '= (255+red+(255+255+255)*0.33)*0.33
              end if
            end if
          end if

          select case as const (i_last and 3)
            case 1   :*pTop = rgb( grn, blu, red )
            case 2   :*pTop = rgb( blu, red, grn )
            case else:*pTop = rgb( red, grn, blu )
          end select
          *pBottom=*pTop
        else
         *pTop   =0
         *pBottom=0
        end if 
      else
        *pTop   =0
        *pBottom=0
      end if
      pTop   +=1
      pBottom-=1
    next
  next

  if (frames mod 100)=0 then
    if len(inkey) then exit do 
  end if
  #ifdef USE_PTC
    ptc_update pBuffer
  #else
    screenunlock
  #endif
loop
' now time
nTime=Timer()

#ifdef USE_PTC
  ptc_close
#endif

screen 0
' clear key buffer for next sleep
while len(inkey):wend

#ifdef USE_PTC
  print "TinyPTC benchmark"
#else
  print "FBGfx   benchmark"
#endif
print "-----------------------------------"
print "counted " & frames & " frames total"
print "frames per seconds = " & frames/(nTime-sTime)
sleep

[marcov]

V1ctor: wrt debug info, do you replicate type information if you import it into a compilation unit?

IOW does a compilation unit contain all info for all types?

[v1ctor]

marcov: FB only supports include files so, that's not a problem.

------

Well, the gcc emitter was just merged into the trunk, you can updated the SVN sources now.

To use it, set an environment variable GCC pointing to your gcc executable (full path + file name).

There are a couple of warning messages that gcc will show, that can be fixed later.

I didn't run the test suite against it (the gcc emitter) yet, neither tried to rebuilt the compiler using it, so, try to build your application that builds fine with 0.21 using it (just add the "-gen gcc" option) and see what happens ;).

[jcfuller]

Well, the gcc emitter was just merged into the trunk, you can updated the SVN sources now.

To use it, set an environment variable GCC pointing to your gcc executable (full path + file name).

There are a couple of warning messages that gcc will show, that can be fixed later.

I didn't run the test suite against it (the gcc emitter) yet, neither tried to rebuilt the compiler using it, so, try to build your application that builds fine with 0.21 using it (just add the "-gen gcc" option) and see what happens ;).

I updated fbc on ubuntu and tested a couple of examples and all worked fine. I did not set an environment variable.
I ran first with -R -c -gen gcc so I could see the files created and all looked fine with a *.c instead of an *.asm file.

James