The MASM Forum

General => The Laboratory => Topic started by: habran on June 11, 2012, 09:30:45 AM

Title: .for, .endfor MACROS
Post by: habran on June 11, 2012, 09:30:45 AM
Hi there,
I am back with these macros and this time with the vengeance
this time I can assure you that they work as expected
take a look at this:
Code: [Select]
xmemcpy PROC USES edi esi dest:DWORD,src:DWORD,count:DWORD
  mov eax,dest
  .if (src != eax)
    .if (count) 
      .for (esi=src,edi=dest,ecx=count:ecx:al=[esi],[edi]=al,esi++,edi++,ecx--)
      .endfor
    .endif
  .endif
  mov eax,dest
 ret
xmemcpy ENDP
;--------------------------------------------------
xmemcpy:
004011ee 55              push    ebp
004011ef 8bec            mov     ebp,esp
004011f1 57              push    edi
004011f2 56              push    esi
004011f3 8b4508          mov     eax,dword ptr [ebp+8]
004011f6 39450c          cmp     dword ptr [ebp+0Ch],eax
004011f9 7424            je      xmemcpy+0x31 (0040121f)
004011fb 837d1000        cmp     dword ptr [ebp+10h],0
004011ff 741e            je      xmemcpy+0x31 (0040121f)
00401201 8b750c          mov     esi,dword ptr [ebp+0Ch]
00401204 8b7d08          mov     edi,dword ptr [ebp+8]
00401207 8b4d10          mov     ecx,dword ptr [ebp+10h]
0040120a eb0d            jmp     xmemcpy+0x2b (00401219)
0040120c 8a06            mov     al,byte ptr [esi]
0040120e 8807            mov     byte ptr [edi],al
00401210 83c601          add     esi,1
00401213 83c701          add     edi,1
00401216 83e901          sub     ecx,1
00401219 23c9            and     ecx,ecx
0040121b 7402            je      xmemcpy+0x31 (0040121f)
0040121d ebed            jmp     xmemcpy+0x1e (0040120c)
0040121f 8b4508          mov     eax,dword ptr [ebp+8]
00401222 5e              pop     esi
00401223 5f              pop     edi
00401224 c9              leave
00401225 c20c00          ret     0Ch
;--------------------------------------------------
;and this would be in x64:
xmemcpy PROC FRAME  dest:QWORD,src :QWORD, count:UINT_PTR
  .if (rcx != rdx)
    .if (r8) 
      .for (:r8:al=[rdx],[rcx]=al,rcx++,rdx++,r8--)
      .endfor
    .endif
  .endif
  mov rax,dest
  ret
xmemcpy ENDP
;--------------------------------------------------
xmemcpy PROC FRAME  dest:QWORD,src :QWORD, count:UINT_PTR
0000000000493496  mov         qword ptr [rsp+8],rcx
000000000049349B  push        rbp 
000000000049349C  mov         rbp,rsp
000000000049349F  cmp         rcx,rdx
00000000004934A2  je          xmemcpy+25h (4934BBh)
00000000004934A4  and         r8,r8
00000000004934A7  je          xmemcpy+25h (4934BBh)
00000000004934A9  mov         al,byte ptr [rdx]
00000000004934AB  mov         byte ptr [rcx],al
00000000004934AD  add         rcx,1
00000000004934B1  add         rdx,1
00000000004934B5  sub         r8,1
00000000004934B9  jmp         xmemcpy+13h (4934A9h)
00000000004934BB  mov         rax,qword ptr [rbp+10h]
00000000004934BF  add         rsp,0
00000000004934C3  pop         rbp 
00000000004934C4  ret             
xmemcpy ENDP
;--------------------------------------------------

and this time I don't care if you like it or not because I am happy with it



 

 
   
Title: Re: .for, .endfor MACROS
Post by: dedndave on June 11, 2012, 10:25:49 AM
we're happy if you're happy   :biggrin:
Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 10:50:42 AM
speak for yourself!!! ;)
Title: Re: .for, .endfor MACROS
Post by: dedndave on June 11, 2012, 10:56:58 AM
and this time I don't care if you like it or not because I am happy with it

what kind of response do you expect ?
i might make suggestions, but it seems you aren't interested
Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 10:58:35 AM
I want you to say that it is a masterpiece and that you love it
Title: Re: .for, .endfor MACROS
Post by: dedndave on June 11, 2012, 11:01:49 AM
i am not crazy about the loop structure in the expanded code
but - the macro looks nice and clean   :biggrin:
Title: Re: .for, .endfor MACROS
Post by: dedndave on June 11, 2012, 11:31:41 AM
i prefer to write code "old style"
i am not too fond of .if/.else or .while/.endw stuff
but - the members don't like it when i write code that way - lol

so - to make code that i like - and that makes them happy,
i have had to learn how the statements work - in reverse
in other words, i write some code, disassemble it, and learn what code is generated
then - i write code they like - that more or less does what i wanted, to begin with

here is a good example
many members will write a message loop like this
Code: [Select]
    .while TRUE
        INVOKE  GetMessage,ebp,edi,edi,edi
    .break .if !eax
        INVOKE  TranslateMessage,ebp
        INVOKE  DispatchMessage,ebp
    .endw
it looks nice - but it generates some fugly code

here is how i prefer to write it
Code: [Select]
        jmp short MLoop1

MLoop0: INVOKE  Translate Message,ebp
        INVOKE  DispatchMessage,ebp

MLoop1: INVOKE  GetMessage,ebp,edi,edi,edi
        inc     eax
        shr     eax,1
        jnz     MLoop0
it might not be pretty, but it is efficient

well - to appease the others - and get what i want, i found a comprimise   :biggrin:
Code: [Select]
    jmp short mEntry

    .while !(ZERO?)
        INVOKE  TranslateMessage,ebp
        INVOKE  DispatchMessage,ebp

mEntry: INVOKE  GetMessage,ebp,edi,edi,edi
        inc     eax                                     ;exit only if 0 or -1
        shr     eax,1
    .endw
the code is amost what i want
it generates an extra JMP that is not executed - but, oh well
Title: Re: .for, .endfor MACROS
Post by: qWord on June 11, 2012, 11:34:58 AM
well - to appease the others - and get what i want, i found a comprimise   :biggrin:
Code: [Select]
    jmp short mEntry

    .while !(ZERO?)
        INVOKE  TranslateMessage,ebp
        INVOKE  DispatchMessage,ebp

mEntry: INVOKE  GetMessage,ebp,edi,edi,edi
        inc     eax                                     ;exit only if 0 or -1
        shr     eax,1
    .endw

the code is amost what i want
it generates an extra JMP that is not executed - but, oh well
OMG  :biggrin:
Title: Re: .for, .endfor MACROS
Post by: dedndave on June 11, 2012, 11:38:01 AM
(http://l.yimg.com/us.yimg.com/i/mesg/emoticons7/24.gif)
qWord is great !!!

ok - maybe that's not the best example - lol
but - the idea is - i had to learn how to use the loop structure in reverse
Title: Re: .for, .endfor MACROS
Post by: jj2007 on June 11, 2012, 12:49:11 PM
I want you to say that it is a masterpiece and that you love it

It is a masterpiece but it looks like C :(
Code: [Select]
      .for (esi=src,edi=dest,ecx=count:ecx:al=[esi],[edi]=al,esi++,edi++,ecx--)
      .endfor
i am a BASIC guy :biggrin:
Code: [Select]
For_ ebx=0 To eax-1
nop
Next
Title: Re: .for, .endfor MACROS
Post by: hutch-- on June 11, 2012, 01:11:07 PM
habran,

I had a look at your macros and you have done some good work there.

i wonder if we can do something like this ?



IF 0  ; ¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤
                      Build this template with "CONSOLE ASSEMBLE AND LINK"
ENDIF ; ¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤

    include \masm32\include\masm32rt.inc

    .data?
      value dd ?

    .data
      item dd 0

    .code

start:
   
; ¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤

    call main
    inkey
    exit

; ¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤

main proc

    LOCAL eax_  :DWORD
    LOCAL ebx_  :DWORD
    LOCAL ecx_  :DWORD
    LOCAL edx_  :DWORD
    LOCAL cntr  :DWORD

    mov cntr, 1

    mov eax, -10
    mov ebx, -11
    mov ecx, -12
    mov edx, -13

  lbl0:

    add eax, 1
    add ebx, 1
    add ecx, 1
    add edx, 1

    .if eax==1 || ebx == 2 || ecx == 3 || edx == 4
      mov eax_, eax
      mov ebx_, ebx
      mov ecx_, ecx
      mov edx_, edx
      pushad
      print ustr$(eax_),13,10
      print ustr$(ebx_),13,10
      print ustr$(ecx_),13,10
      print ustr$(edx_),13,10
      popad
    .else
      pushad
      print "iteration "
      print ustr$(cntr),13,10
      add cntr, 1
      popad
      jmp lbl0
    .endif

    ret

main endp

; ¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤

end start


JJ,  :t It really WAS a typo.
Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 03:11:21 PM
thanks huch

if we put this statement in brackets 
.if eax==1 || ebx == 2 || ecx == 3 || edx == 4

.if (eax==1 || ebx == 2 || ecx == 3 || edx == 4)
it will work fine
 the rest is a chicken shit

best regards


Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 03:17:49 PM
jj2007

c like forloop can not look like BASIC FOR loop

but thank you for appreciating it

thanks to all of you for looking in to it
Title: Re: .for, .endfor MACROS
Post by: hutch-- on June 11, 2012, 03:29:24 PM
 :biggrin:

> it will work fine

It works fine without the brackets.  :P
Title: Re: .for, .endfor MACROS
Post by: Ghandi on June 11, 2012, 05:57:29 PM
Is there any gain to be had anymore from separating small memory copy operations from larger ones and using different methods such as SSE2, FPU, loading and incrementing pointers like in these macro examples or even the older REP MOVSx operands?

If so, you should be able to add conditional statements to only assemble the required method for the size given as a macro parameter shouldn't you?

Interesting, i'm going to look at this myself for curiousity's sake.

HR,
Ghandi
Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 09:10:47 PM
Mea culpa, huch :icon_mrgreen:
I was in hurry and still my mind was in the .for macro
however, it looks much better when in brackets, you have to admit that
So, what is actually that you are concerned with in that example?
Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 09:53:14 PM
dedndave,
>i prefer to write code "old style"
I appreciate your opinion
however, sometimes readability of program is more important then speed
ml64 is out of game because of the "old style"

best regards
 
Title: Re: .for, .endfor MACROS
Post by: jj2007 on June 11, 2012, 10:08:03 PM
however, it looks much better when in brackets, you have to admit that

There is room for improvement then:
Code: [Select]
.if eax
nop
.endif

can become
Code: [Select]
.if (eax!=NULL);
;{
nop;
;}
.endif
;)
Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 10:28:04 PM
 :eusa_clap: jj2007, I like your cinizam
I take it as a great sense of humor
that is a masterpiece jock,  I see it first time in my life
you should publish it somewhere

BTW
I used to write in BSIC while using commodore 64 but just to transfer machine code to assembler or opposite
I like C more than BASIC but my favorite is ASM, nothing can compensate for it
Title: Re: .for, .endfor MACROS
Post by: qWord on June 11, 2012, 10:46:55 PM
For-loops are fine and I’ve also had the idea to create such a macro, but finial I end up that the resulting syntax doesn't look as nice as in C. Currently I can only see a change if it is directly implemented in the assembler:  jWasm.
For your macros, I see the problem of the used none-ASCII-characters (<<,>>), which are only visible, if the editor use the correct code page.
Title: Re: .for, .endfor MACROS
Post by: habran on June 11, 2012, 10:59:24 PM
qWord, in this case you can use LT, GT, EQ, LE, NE, GE

>directly implemented in the assembler:  jWasm.

I have worked on it and already build the skeleton of it
but I did not have enough time to study the code necessary to translate operators into code

now I am planing to look at it
Title: Re: .for, .endfor MACROS
Post by: jj2007 on June 12, 2012, 01:03:00 AM
I used to write in BASIC while using commodore 64 but just to transfer machine code to assembler or opposite
I like C more than BASIC but my favorite is ASM, nothing can compensate for it

I see we are on the same wavelength :biggrin:
My "coding career" was always a mix of Basic (i.e. Gfa on the Atari ST - almost a Pascal dialect) and assembler (68k, later x86). So I am probably just too old for C - and anyway, in Olly they look remarkably similar ;-)

Excuse my ignorance: What is the syntax for an extra simple loop as shown below?
include \masm32\MasmBasic\MasmBasic.inc   ; download (http://masm32.com/board/index.php?topic=94.0)
   Init
   PrintLine "MasmBasic For...Next loop:"
   mov eax, 5
   For_ ecx=0 To eax-1
      Print Str$("ecx=%i ", ecx)
   Next
   PrintLine CrLf$, "Now Habran's macro:"
   ; .FOR (
   Inkey "ok"
   Exit
end start

Output:
MasmBasic For...Next loop:
ecx=0 ecx=1 ecx=2 ecx=3 ecx=4
Now Habran's macro:
ok


P.S.: Any news from JWasm? 2.07 pre is fine for most purposes but there are minor quirks to fix...
Title: Re: .for, .endfor MACROS
Post by: habran on June 12, 2012, 07:18:24 AM
jj2007
wavelength? you mean we remember things happened deep in the last millennium?
no mobile, no internet, no computer no nothing except having sex 28 hour a day!!! 8)

try this because I did not try it:
Code: [Select]
Init
   PrintLine "MasmBasic For...Next loop:"
   .for (eax=5,ecx=0:SDWORD PTR eax GE 0:eax--)
      Print Str$("ecx=%i ", ecx)
      PrintLine CrLf$, "Now Habran's macro:"
      ; .FOR (
      Inkey "ok"
   .endfor
   Exit
end start

best regards neighbor

P.S.: "0" (JWasm news)
 
Title: Re: .for, .endfor MACROS
Post by: hutch-- on June 12, 2012, 10:36:38 AM
There is nothing wrong with the analog world, it keeps running when the digital world's batteries go flat.  :biggrin:
Title: Re: .for, .endfor MACROS
Post by: dedndave on June 12, 2012, 01:01:25 PM
shouldn't ".endfor" be ".next"   :lol:
Title: Re: .for, .endfor MACROS
Post by: jj2007 on June 12, 2012, 01:17:47 PM
wavelength? you mean we remember things happened deep in the last millennium?
no mobile, no internet, no computer no nothing except having sex 28 hour a day!!! 8)

Hutch, alarm!! We have a Martian on board - earthlings had only 24 hours available for this ;-)

Tried your code but no luck, it throws lots of error messages, see attachment...

include \masm32\MasmBasic\MasmBasic.inc   ; download (http://masm32.com/board/index.php?topic=94.0)
include FOR_M.inc   ; slightly modified
   Init

   PrintLine "MasmBasic For...Next loop:"
   mov eax, 5
   For_ ecx=0 To eax-1
      Print Str$("ecx=%i ", ecx)
   Next

   PrintLine CrLf$, "Now Habran's macro:"
   .for (eax=5:ecx=0:SDWORD PTR eax GE 0:eax--)
      nop
   .endfor   

   Inkey "ok"
   Exit
end start
Title: Re: .for, .endfor MACROS
Post by: habran on June 12, 2012, 02:58:14 PM
jj2007
sorry it was supposed to be with comma not with column

like this: .for (eax=5,ecx=0:SDWORD PTR eax GE 0:eax--)

and I am not from Mars, I am your neighbor fro former Yugoslavia
but I am already 20 years in South Australia
Title: Re: .for, .endfor MACROS
Post by: habran on June 12, 2012, 03:00:13 PM
 :biggrin: Yeah hutch: analog sex 8)
Title: Re: .for, .endfor MACROS
Post by: jj2007 on June 12, 2012, 03:42:05 PM
jj2007
sorry it was supposed to be with comma not with column

like this: .for (eax=5,ecx=0:SDWORD PTR eax GE 0:eax--)

assembles fine now but doesn't do anything ::)
But I got it working in the end:

include \masm32\MasmBasic\MasmBasic.inc   ; download (http://masm32.com/board/index.php?topic=94.0)
include FOR_M.inc   ; slightly modified
   Init

   PrintLine "MasmBasic For...Next loop:"
   mov eax, 5
   For_ ecx=0 To eax-1
      Print Str$("ecx=%i ", ecx)
   Next

   PrintLine CrLf$, CrLf$, "Now Habran's macro: "
;   mov eax, 4
   .for (ecx=0 :SDWORD PTR ecx GE 4 : ecx++)
      Print Str$("ecx=%i ", ecx)
   .endfor   
   Inkey CrLf$, "ok"
   Exit
end start

Output:
MasmBasic For...Next loop:
ecx=0 ecx=1 ecx=2 ecx=3 ecx=4

Now Habran's macro:
ecx=0 ecx=1 ecx=2 ecx=3 ecx=4


Two minor issues:
   mov eax, 4
   .for (ecx=0 : SDWORD PTR ecx GE eax : ecx++)   ; reg32 instead of immediate: endless loop, eax gets trashed somewhere

   .for (ecx=0 : SDWORD PTR ecx GE 4 : ecx++)   ; blank before SDWORD: assembly hangs


Quote
and I am not from Mars, I am your neighbor fro former Yugoslavia
but I am already 20 years in South Australia

cosmically speaking, Mars is a close neighbour, too :biggrin:
by the way, if I remember well, even a century back that 24-hour activity was already digital ;)
Title: Re: .for, .endfor MACROS
Post by: habran on June 12, 2012, 05:10:39 PM
jj2007

I will look at it tonigh :eusa_clap:

are you sure that we are from the same century? :lol:
Title: Re: .for, .endfor MACROS
Post by: jj2007 on June 12, 2012, 05:52:08 PM
are you sure that we are from the same century? :lol:

Pretty sure. I am uncertain about the location though - Mars is the only planet with a more than 24 hours day ::)
Title: Re: .for, .endfor MACROS
Post by: habran on June 12, 2012, 07:35:50 PM
OK, when I said 28 hours a day I meant more than I needed
that means that we exaggerated in a pleasure
And for your information, Mars is not inhabited!!! (Except Arnold Schwarzenegger :badgrin:)
Not any more, they died of boredom because they did not have internet
If you don't believe me ask other members on this board

best regards
Title: Re: .for, .endfor MACROS
Post by: FORTRANS on June 12, 2012, 11:58:52 PM
Hi,

Quote
Mars is the only planet with a more than 24 hours day

   Only if you ignore Mercury, Venus, and optionally Pluto.

Cheers,

Steve N.
Title: Re: .for, .endfor MACROS
Post by: dedndave on June 13, 2012, 12:00:12 AM
i think Maria Shriver is from Pluto   :redface:
Title: Re: .for, .endfor MACROS
Post by: jj2007 on June 13, 2012, 12:15:00 AM
Quote
Mars is the only planet with a more than 24 hours day

   Only if you ignore Mercury, Venus, and optionally Pluto.


Steve,
We only considered those which are appropriate for digital activity (hey, did anybody notice we are in the Lab here? we'll soon be banned :bgrin:)
Title: Re: .for, .endfor MACROS
Post by: FORTRANS on June 13, 2012, 12:27:38 AM
Hi jj,

   Well, okay.  I'll quiet up after mentioning that Messenger
is now in orbit around Mercury.

Cheers,

Steve N.
Title: Re: .for, .endfor MACROS
Post by: habran on June 13, 2012, 01:10:30 AM
thank you jj2007 for finding the bug in the macro :t
it was a proof that I am just a human (and not a Martian) :eusa_boohoo:
the bug was there because I was lazy to type so I copied and pasted the same routine but missed
to change 1 number
it was here:
Code: [Select]
:Lop3
  cspace TEXTEQU @SubStr(<params>,temp,1)
   IFIDN cspace,< >
   temp = temp+1
   space1 = space1+1
   goto Lop2               ;bug was here, it supposed to be goto Lop3
   ENDIF                           

I have fixed it an upload it in the beginning of this tread
and now it works like a charm :eusa_clap:

 thanks again
Title: Re: .for, .endfor MACROS
Post by: mywan on July 17, 2012, 01:48:24 PM
here is how i prefer to write it
Code: [Select]
        jmp short MLoop1

MLoop0: INVOKE  Translate Message,ebp
        INVOKE  DispatchMessage,ebp

MLoop1: INVOKE  GetMessage,ebp,edi,edi,edi
        inc     eax
        shr     eax,1
        jnz     MLoop0
it might not be pretty, but it is efficient

Not sure what's not pretty about it. One of the things that worried me when deciding to take on learning asm was loops and switches. To me the quoted snippet was the easiest to comprehend with the least effort of the 3 you provided. I have no issues at all with higher level syntax styles when it adds clarity, ease of use, or comprehensibility. Even paying some mild overhead cost for it is not out of the question in most circumstances. I'm just not personally seeing any of that in the examples given.

I hope if you contribute to my questions in the future you can offer your "old style" perspective as well. It's valuable even when alternative approaches are chosen in a given instance.
Title: Re: .for, .endfor MACROS
Post by: dedndave on July 17, 2012, 08:55:11 PM
i feel better, now

some of the guys in here can't read code unless it is surrounded by ".IF/.ENDIF"   :lol:
i hate that stuff, but i want to use it so that the newbies are accustomed to it
Title: Re: .for, .endfor MACROS
Post by: Ryan on July 18, 2012, 01:59:40 AM
Dave, any reason why you use inc and shr instead of cmp?
Title: Re: .for, .endfor MACROS
Post by: jj2007 on July 18, 2012, 02:02:16 AM
Dave, any reason why you use inc and shr instead of cmp?

Loop exits for msg=0 and msg=-1 ;-)
Title: Re: .for, .endfor MACROS
Post by: dedndave on July 18, 2012, 02:14:27 AM
to expand on what Jochen said...

by using INC and SHR, we get to use only 1 branch instruction

INC reg32 is a single-byte instruction
SHR reg32,1 is a two-byte instruction
JNZ by a reverse distance of 128 bytes or less is 2 bytes
Code: [Select]
        inc     eax
        shr     eax,1
        jnz     MLoop0
5 bytes total

we want to exit the program if the returned value is either 0 or -1
for all other values, we want to loop
Title: Re: .for, .endfor MACROS
Post by: Ryan on July 18, 2012, 02:34:46 AM
Yeah, I guess
Code: [Select]
.break .if !eax would fail to exit if -1 is returned.
Title: Re: .for, .endfor MACROS
Post by: jj2007 on July 18, 2012, 03:10:51 AM
Good guess :biggrin:

Dave's code is one cycle faster and 4 bytes shorter.
Title: Re: .for, .endfor MACROS
Post by: Ryan on July 18, 2012, 03:25:22 AM
The irony is the template that Microsoft uses for new C++ programs doesn't account for -1.  I wish they would elaborate on what exactly would cause it to fail.  They mention as examples an invalid hWnd or an invalid message pointer.  Passing hWnd would overlook WM_QUIT, so that's definitely not recommended in a primary message loop.  If someone passes an invalid message pointer, it would certainly be caught on the first run.
Title: Re: .for, .endfor MACROS
Post by: qWord on July 18, 2012, 03:27:01 AM
optimizing loops with blocking functions is like makeup for dogs  :biggrin:
Title: Re: .for, .endfor MACROS
Post by: mywan on July 18, 2012, 03:55:56 AM
i feel better, now

some of the guys in here can't read code unless it is surrounded by ".IF/.ENDIF"   :lol:
i hate that stuff, but i want to use it so that the newbies are accustomed to it
While trying to learn from reading source basic ".IF/.ENDIF" is not a problem. But when it has a lots of nested ".ELSE/.ELSEIF" I find myself messing with the white space to try and make sense of it. It seems a lot cleaner to me to see jnz. That tells me immediately where it is going and why without tracking back up through the page to see what other conditionals are contained in ".IF/.ELSEIF/.IF/.ELSEIF/.ENDIF/.ELSEIF/.ENDIF". No matter how many times I get it straight I still have to rework the logical blocks through my mind when I go back.

Edit: Somehow double quoted the same quote.
Title: Re: .for, .endfor MACROS
Post by: dedndave on July 18, 2012, 04:06:20 AM
mywan - my point, exactly   :t

qword - what blocking function ?
Quote
optimizing loops with blocking functions is like makeup for dogs
some dogs are just plain ugly - lol
you and Hutch can both write bloatware for MS
my approach has always been:
1) if speed matters, make it fast
2) if speed doesn't matter, make it small
in general, i try to develop habits that are aligned with that approach
so - even if it isn't critical, i am developing the habits that i aim for

Ryan -
Quote
I wish they would elaborate on what exactly would cause it to fail.
a sentiment that has been expressed by more than one other member   :biggrin:

as for the errors that MSDN says may be encountered
1) if it has an incorrect hWnd, it should never get into the queue - DOH !
2) invalid message pointer  :dazzled:
again - if there is an invalid pointer in a message parm, the message should return an error

at any rate, i have never seen this error occur
and - if it did, there may be better ways to handle it than exiting the program
for example - you could just pull the offending message out of the queue and move on
but - i haven't seen anyone's code do ths   :P
i figure exiting the program is safer than dispatching an erroneous message
Title: Re: .for, .endfor MACROS
Post by: qWord on July 18, 2012, 04:34:10 AM
qword - what blocking function ?
functions that get into a wait-state (GetMessage()).

IMO it is most important to make code readable and understandable. Optimize code is worthless, if no one understand it. Even, if you do not share the code with others, the question is if you will understand the code in some years. This “spaghetti”-style (no .if/.while,...) requires a lot more comments and I don’t think these will increase the transparency in any case.
My strategy is:  if (Speed critically) then (optimize as much as possible) else (make code readable)
Title: Re: .for, .endfor MACROS
Post by: jj2007 on July 18, 2012, 04:41:36 AM
It's not speed, of course - but saving 4 bytes is always nice ;)

the question is if you will understand the code in some years. This “spaghetti”-style (no .if/.while,...) requires a lot more comments

True. It doesn't matter when you have 20 lines of code, but for anything more complex the HLL constructs help a lot - there is a reason why GOTO was banned from modern languages.

I use both versions, actually, and I couldn't really explain why I sometimes prefer a jnz...
But I definitely need colour:
   push ebx
   push esi
   ... 20 lines of code
   pop esi
   ... 20 lines of code
   pop ebx

Dave, under which circumstances can you produce -1? My windows always exit with zero...
Title: Re: .for, .endfor MACROS
Post by: mywan on July 18, 2012, 05:24:13 AM
functions that get into a wait-state (GetMessage()).

IMO it is most important to make code readable and understandable. Optimize code is worthless, if no one understand it. Even, if you do not share the code with others, the question is if you will understand the code in some years. This “spaghetti”-style (no .if/.while,...) requires a lot more comments and I don’t think these will increase the transparency in any case.
My strategy is:  if (Speed critically) then (optimize as much as possible) else (make code readable)
As someone just getting started with asm I'm finding the "no .if/.while,..." structures quicker to comprehend than with them, even after already figuring out the logical blocks of the ".IF/.THEN", and this is other peoples uncommented code.  Hence the notion that it will create issues for me to understand my own code after some years seems unlikely. Absolutely love the "invoke" for API calls though. I am absolute not against macros and higher level constructs, but the assumption that these constructs always make code more readable just doesn't compute in some minds.

Coming from a nooby position the most valuable _missing_ feature is context sensitive help files. When I can't remember which jump "jnz", etc., refers to my editor is perfectly capable of going directly to "\MASM32\HELP\help.chm::/jnz.htm" simply by hitting F1 with the cursor over "jnz". Neither are the help files searchable. This for me, in both cases, is a lot worse than any lack of comments. It is, imo, well worth creating a project just to remedy this.
Title: Re: .for, .endfor MACROS
Post by: jj2007 on July 18, 2012, 05:38:23 AM
the most valuable _missing_ feature is context sensitive help files. When I can't remember which jump "jnz", etc., refers to my editor is perfectly capable of going directly to "\MASM32\HELP\help.chm::/jnz.htm" simply by hitting F1 with the cursor over "jnz".

Good idea, but Microsoft makes it very difficult. In my own editor, I have chained two help files, so that F1 on WinByTitle opens the MasmBasic help file, while F1 on SetForegroundWindow opens the page in Win32.hlp (because it can't find SetForegroundWindow in the MasmBasic help). The same with *.chm files is a lot more difficult. If the bloody chm does not contain an exact match SetForegroundWindow.htm, you just get a boring "page not found" - and most of the time it's not even searchable. Stupid, very stupid :(
Title: Re: .for, .endfor MACROS
Post by: mywan on July 18, 2012, 07:39:06 AM
I just created a thread asking about interest in a help file project like this. I have seen it done very effectively with some scripting languages. Initially it needs to be done as <keyword>.htm all in the same folder. From there it can easily be compiled into an effective integratable .chm file. Certain standards should be set on the structure of the .htm files.