Index
> Main > tree viewGoto page 1, 2 Next |
| Author |
|
|
typedef 21 Mar 2011, 20:12
Are you asking or saying/teaching/informing?
|
|||
21 Mar 2011, 20:12 |
|
|
edfed 21 Mar 2011, 20:31
both.
try to inform, and launch some development around trees in asm. |
|||
|
21 Mar 2011, 20:31 |
|
|
typedef 21 Mar 2011, 20:48
Hmm Interesting. The reason why I was asking was because I had an I dea to contribute, but I think you may have already thought about it.
I was thingking that you make an event callbacks engine that will dispatch the events to registered listeners. Like when you click on a DOT, the callback engine will dispatch what type of click it was {Double Or Single Click or maybe the middle button, path of the tree, parent branch ( With a pointer to it's children branche(s)) } With the location of the cursor, if needed. So let's say I made an app with a tree. So I would register a routine that would receive the dispatched events. Ex. Code: let's say the event dispatcher calls the registered routines with the following parameters BX points to tree .The one I created DI points to tree data sctructure. AX points to eventId MyTreeEventRoutine: ; proc PUSH BP SP DI BX cmp ax,TREE_CLICKED je .clicked cmp ax,TREE_COLLAPSED je .collapsed ... .clicked: DI contains what we want [DI+4] which button was clicked ( Left, Middle, Right) [DI+8] High byte (point: X) Low byte(Point: Y) etc jmp .loopCode POP ... ret endp You get what I mean ?  |
|||
|
21 Mar 2011, 20:48 |
|
|
edfed 21 Mar 2011, 22:12
i get it, it have a name.
fool. with trees, the dimension of programming changes a lot. instead of having sequantial code with branches like you stated, you have multiple trees of action, items, initialisations, options, etc.. it looks a little like the windows registry, but is direct, no extra layer, direct asm execution behind the items. tree programming is like a virtual machine. instead of sequential CPU, you have item CPU. thaat will travel in a tree instead of a sequence. from outside, it is the same as any programming langage. but the difference is data vs code. |
|||
|
21 Mar 2011, 22:12 |
|
|
typedef 22 Mar 2011, 01:01
Sounds like a lot of recursive?
Anyways, I'm yet to see nested recursive functions in Assembly though.... Can you make the nested fibonacci recursive function that calls itself in another fibonacci function......Heheheheh.......right now.... I'm not demanding  |
|||
|
22 Mar 2011, 01:01 |
|
|
typedef 22 Mar 2011, 01:02
BTW, does your tree have a fixed length and size ?
|
|||
|
22 Mar 2011, 01:02 |
|
|
edfed 22 Mar 2011, 01:11
recursion is possible, but should be limited with a counter.
@@:call @b is an example of impossible recursion. mov cx,10 @@: dec ecx je @f call @b is possible reccursion.. a tree by definition have not a fixed size, nor lengh. each tilme you add a node, you can add nodes inside, etc, etc... |
|||
|
22 Mar 2011, 01:11 |
|
|
typedef 22 Mar 2011, 01:17
Hmmm.....sorry, this is about your tree but I have some questions.
So which is faster.... @@: ... call @b or @@: ... jmp @b or does it depend on what goes in between ?  |
|||
|
22 Mar 2011, 01:17 |
|
|
edfed 22 Mar 2011, 01:38
jmp @b is a loop
call @b is a recursion. when programming something like that, don't care about what is faster, but what is better. |
|||
|
22 Mar 2011, 01:38 |
|
|
typedef 22 Mar 2011, 02:28
edfed wrote: when programming something like that, don't care about what is faster, but what is better. point well taken   |
|||
|
22 Mar 2011, 02:28 |
|
|
iic2 24 Mar 2011, 02:32
Only a foo could find this one ... but it's in C++.
http://www.bogotobogo.com/cplusplus/binarytree.html .... .... output Code: size = 9 max depth = 4 min depth = 2 This tree is not balanced! Min value = A Max value = I Min value of subtree B as a root is A Max value of subtree B as a root is E In Order Successor of B is C In Order Successor of E is F In Order Successor of I is None In Order Predecessor of B is A In Order Predecessor of E is D In Order Predecessor of I is H The lowest common ancestor of A and C is B The lowest common ancestor of E and H is F increasing sort order A B C D E F G H I post order A C E D B H I G F pre order F B A D C E G I H D is in the tree M is not in the tree printing paths ... F B A F B D C F B D E F G I H 5th maximum data is E New array: A B C D E F G H I 12 subtree: 1 13 subtree: 1 14 subtree: 0 23 subtree: 0 32 subtree: 1 printing with Breadthfirst traversal E B G A C F H D I Press any key to continue . . . |
|||
|
24 Mar 2011, 02:32 |
|
|
edfed 24 Mar 2011, 14:10
a little advance in the tree viewing.
now, i will make multiples parts: data structure methods (folder, file, lba, sectors, header) tree exploration (down1level, up1level, current path , show content, etc...) graphics display (container, wire, name, properties...) for the moment, it gives that: 
one very hard thing about tree management is when it is intended to be a file system, then, stocked on drives, with the problems by usage of the drives (bad sectors, fragmentation, variable sector size,...) but i think there is a possible layer for this: the current code is a little more structurised, explaining why the view is more structurised. Code:
org 100h
mov ax,0a000h
mov es,ax
mov ax,13h
int 10h
mov si,tree
call caller
ret
tree: dw node,@f-$-2,\
.F,.D,.1,.2,.3,.4,.5,.6,.1
@@:
.1: dw node,@f-$-2,\
.7,.8,.9,.A,.B,.C,.D
@@:
.2: dw node,@f-$-2,\
.E,.F,.1,.3
@@:
.3: dw node,@f-$-2,\
.1,.A,.B,.C,.D,.1
@@:
.4: dw node,@f-$-2,\
.7,.8,.9,.A,.B,.C,.D
@@:
.5: dw node,@f-$-2
@@:
.6: dw item
.7: dw item
.8: dw item
.9: dw item
.A: dw item
.B: dw item
.C: dw item
.D: dw item
.E: dw item
.F: dw item
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
caller:
push ax
cmp si,0
je @f
mov ax,[si]
cmp ax,0
je @f
call ax
@@:
pop ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
node:
push ax bx cx [x] [y]
mov [c],1
call spot
inc [y]
mov cx,[si+2]
shr cx,1
je .end
mov bx,0
add [x],lengh
@@:
push si
add [y],3
mov [c],35
call horizontal
mov si,[si+4+bx]
call caller
pop si
add bx,2
loop @b
.end:
pop ax [x] cx bx
push [y]
sub ax,[y]
neg ax
mov [yl],ax
mov [c],33
call vertical
pop [y] ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
item:
mov [c],4
call spot
inc [y]
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
firstpixel:
mov ax,[x]
mov bx,[y]
shl bx,2
add bx,[y]
shl bx,6
add bx,ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
putpixel:
push ax bx cx
call firstpixel
mov cl,[c]
mov [es:bx],cl
pop cx bx ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
spot:
push ax bx cx
call firstpixel
mov cl,[c]
; mov [es:bx],cl
mov [es:bx+1],cl
mov [es:bx-1],cl
mov [es:bx+1+320],cl
mov [es:bx-1+320],cl
mov [es:bx+1-320],cl
mov [es:bx-1-320],cl
mov [es:bx+320],cl
mov [es:bx-320],cl
pop cx bx ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
plot:
push ax bx cx
call firstpixel
mov cl,[c]
mov [es:bx],cl
mov [es:bx+1],cl
mov [es:bx-1],cl
mov [es:bx+320],cl
mov [es:bx-320],cl
pop cx bx ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vertical:
;yl
push ax bx cx
call firstpixel
mov dl,[c]
mov cx,[yl]
cmp cx,0
jl .end
@@:
sub bx,320
mov [es:bx],dl
loop @b
.end:
pop cx bx ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
horizontal:
;yl
push ax bx cx
call firstpixel
mov dl,[c]
mov cx,[xl]
cmp cx,0
jl .end
@@:
sub bx,1
mov [es:bx],dl
loop @b
.end:
pop cx bx ax
ret
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
x dw 0
y dw 0
c db 0
yl dw 1
xl dw lengh
lengh = 8
|
|||
|
24 Mar 2011, 14:10 |
|
|
typedef 25 Mar 2011, 00:26
^What video mode are you using ?
|
|||
|
25 Mar 2011, 00:26 |
|
|
revolution 25 Mar 2011, 00:28
typedef wrote: ^What video mode are you using ? Code: mov ax,13h int 10h |
|||
|
25 Mar 2011, 00:28 |
|
|
typedef 26 Mar 2011, 21:49
^Did not see that hheheehe, and I did not have to read since you are here
 |
|||
|
26 Mar 2011, 21:49 |
|
|
typedef 26 Mar 2011, 21:52
Oh, and Btw, I had to add
Code: mov ah,0 int 16h after Code:
call caller
|
|||
|
26 Mar 2011, 21:52 |
|
|
edfed 26 Mar 2011, 23:27
maybe yes, wait a keystroke can let you see the screen if you are not under windows 98.
on my machine, it is not a problem since the console don't close when ret from mode 13h. to close the console, should reswitch in mode 3, and ret. no need of int 20h or else, by testing, it appears that just a ret is enough to quit a .com program. now that i have retested the tree layer with succes, i plan on the file descriptor layer. fdesc will be simple: fdesc: .type dd ? .name dd ? .size dd ? .loc dd ? each entry in a folder will be a pointer to a fdesc. then, folder: ;(is a file) .size dd size of entry list .entry dd ? .entry dd ? etc... entry points to a fdesc inside the folder file fdesc.name points to a asciiz string. fdesc.loc will be the lba of the first sector of the sector list. zones list will be a list like this: .size dd ? .lba dd ? .sectors dd ? .lba dd ? .sectors dd ? .lba dd ? .sectors dd ? .lba dd ? .sectors dd ? each zone (lba, sectors) will be a chunk of the file. there will be the possibility of just one single zone, as there will be possible to have thousand of them, repartited in several sectors. if there are more than one sector for the zones list, the last dword of the sector will contain the lba for the next. file should be recorded as contiguous paged bytes, without extra formating. than the zones list. from the application view, file will be a linear adress space, starting at 0. from the specific uses of files (depends on type), the loading adress will be determined. for example, if fdesc.type = "com", file will be loaded by com function at 100h, induce the creation of the psp, and lanch it... every file types will consist on either a id (just a 32 bits number) or a string (4 bytes string)... for a file system, apparently, there is the need to cut the tree in severals layers, due to the polymorphism of the data structures... and then, the tree layer (the subject of this topic) is closer to the user than the others stated in this post. fdesc, zonelist, bitmap, etc... should be transparent. |
|||
|
26 Mar 2011, 23:27 |
|
|
revolution 26 Mar 2011, 23:37
edfed wrote: no need of int 20h or else, by testing, it appears that just a ret is enough to quit a .com program. |
|||
|
26 Mar 2011, 23:37 |
|
|
edfed 26 Mar 2011, 23:42
it is not the question of one byte, but just because it is easier (and faster) to write ret than int 20h
|
|||
|
26 Mar 2011, 23:42 |
|
| Goto page 1, 2 Next < Last Thread | Next Thread > |
Forum Rules:
|