Write a compiler. Some possibility?

2Player · Post by **2Player** » Sat Dec 17, 2005 2:09 am

I'm thinking. Has any possibility of to write a new programming language and a compiler, using the PureBasic?

It's possible or I'm Crazy?

Shannara · Post by **Shannara** » Sat Dec 17, 2005 8:23 am

You bet, with compilers written in VB, you can pretty easily convert it over to PB ... reference: http://www.kidev.com

Trond · Post by **Trond** » Sat Dec 17, 2005 10:34 am

Sure it's possible, I've got one in front of me (not usable for anything yet but has a lot of skeleton like expression with many operators and constructs for conditional execution), but I don't want to give you the source.

utopiomania · Post by **utopiomania** » Sat Dec 17, 2005 7:09 pm

I don't know about compilers, but what about interpreters? It would require a runtime, and could be used as a base
for a scripting language i suppose.

There's quite a few sources around, and I think it would be nice to have one of them translated to PureBasic.

venom · Post by **venom** » Sat Dec 17, 2005 7:39 pm

I'm kind of working on a virtual machine and assembler thats so far been written in PureBasic, it currently can do calculations and pointer handling.

utopiomania · Post by **utopiomania** » Sat Dec 17, 2005 8:16 pm

To clarify what I said, here's a working Basic Interpreter, submitted to an obfuscated code contest years ago written
in ANSI C. If there are any C programmers here, the first step is to submit it to a preprocessor.

Any taker's? At least about running it through a preprocessor and postiing the result, or PM it to me.

Code: Select all

#define O(b,f,u,s,c,a)b(){int o=f();switch(*p++){X u:_ o s b();X c:_ o a b();
default:p--;_ o;}}
#define t(e,d,_,C)X e:f=fopen(B+d,_);C;fclose(f)
#define U(y,z)while(p=Q(s,y))*p++=z,*p=' '
#define N for(i=0;i<11*R;i++)m[i]&&
#define I "%d %s\n",i,m[i]
#define X ;break;case
#define _ return
#define R 999
typedef char*A;int*C,E[R],L[R],M[R],P[R],l,i,j;char B[R],F[2];A m[12*R],malloc
(),p,q,x,y,z,s,d,f,fopen();A Q(s,o)A s,o;{for(x=s;*x;x++){for(y=x,z=o;*z&&*y==
*z;y++)z++;if(z>o&&!*z)_ x;}_	0;}main(){m[11*R]="E";while(puts("Ok"),gets(B)
)switch(*B){X'R':C=E;l=1;for(i=0;i<R;P[i++]=0);while(l){while(!(s=m[l]))l++;if
(!Q(s,"\"")){U("<>",'#');U("<=",'$');U(">=",'!');}d=B;while(*F=*s){*s=='"'&&j
++;if(j&1||!Q(" \t",F))*d++=*s;s++;}*d--=j=0;if(B[1]!='=')switch(*B){X'E':l=-1
X'R':B[2]!='M'&&(l=*--C)X'I':B[1]=='N'?gets(p=B),P[*d]=S():(*(q=Q(B,"TH"))=0,p
=B+2,S()&&(p=q+4,l=S()-1))X'P':B[5]=='"'?*d=0,puts(B+6):(p=B+5,printf("%d\n",S
()))X'G':p=B+4,B[2]=='S'&&(*C++=l,p++),l=S()-1 X'F':*(q=Q(B,"TO"))=0;p=B+5;P[i
=B[3]]=S();p=q+2;M[i]=S();L[i]=l X'N':++P[*d]<=M[*d]&&(l=L[*d]);}else p=B+2,P[
*B]=S();l++;}X'L':N printf(I)X'N':N free(m[i]),m[i]=0	X'B':_ 0 t('S',5,"w",N
fprintf(f,I))t('O',4,"r",while(fgets(B,R,f))(*Q(B,"\n")=0,G()))X 0:default:G()
;}_ 0;}G(){l=atoi(B);m[l]&&free(m[l]);(p=Q(B," "))?strcpy(m[l]=malloc(strlen(p
)),p+1):(m[l]=0,0);}O(S,J,'=',==,'#',!=)O(J,K,'<',<,'>',>)O(K,V,'$',<=,'!',>=)
O(V,W,'+',+,'-',-)O(W,Y,'*',*,'/',/)Y(){int o;_*p=='-'?p++,-Y():*p>='0'&&*p<=
'9'?strtol(p,&p,0):*p=='('?p++,o=S(),p++,o:P[*p++];}

Trond · Post by **Trond** » Sat Dec 17, 2005 9:37 pm

Interpreters and virtual machines are all crap. They're not fast enough. Sure, you can't expect an interpreter to be fast. Is that an excuse? I sent you a Fiat instead of a Lamborghini you ordered so don't expect it to be fast?

By the way, I can make a working Basic interpreter too (not as obfuscated), but who wants an interpreter?

utopiomania · Post by **utopiomania** » Sat Dec 17, 2005 9:42 pm

but who wants an interpreter?

I do, and I also decides if it's fast enough for my use, not you.

Heathen · Post by **Heathen** » Sat Dec 17, 2005 10:26 pm

Trond wrote:Interpreters and virtual machines are all crap. They're not fast enough. Sure, you can't expect an interpreter to be fast. Is that an excuse? I sent you a Fiat instead of a Lamborghini you ordered so don't expect it to be fast?

By the way, I can make a working Basic interpreter too (not as obfuscated), but who wants an interpreter?

I disagree with you, vms are not always slow. Depending on how much preprocessing is done, they can be very fast.

thefool · Post by **thefool** » Sat Dec 17, 2005 11:07 pm

Trond wrote:Interpreters and virtual machines are all crap. They're not fast enough. Sure, you can't expect an interpreter to be fast. Is that an excuse? I sent you a Fiat instead of a Lamborghini you ordered so don't expect it to be fast?

By the way, I can make a working Basic interpreter too (not as obfuscated), but who wants an interpreter?

heh i bet you. Java IS fast. VERY fast!

Nik · Post by **Nik** » Sat Dec 17, 2005 11:23 pm

But NetBeans IDE is so slow I hated to use it so much that I quit Java but I think this is because Java uses this wierd non standard Windowsystem.

thefool · Post by **thefool** » Sat Dec 17, 2005 11:45 pm

Downloading

MrMat · Post by **MrMat** » Sun Dec 18, 2005 12:08 am

NetBeans uses Swing in which each gadget is effectively hand-drawn. This is slower but fits in with Sun's "write once, run anywhere" mantra because it is independent of the OS it is running on. Another graphical interface library is SWT, as used by Eclipse, which acts as a wrapper around the native OS GUI routines, just like PB does and is thus faster and feels more "natural". On the downside this approach requires updating the library for each OS it is to run on. It is up to the author what they use.

Killswitch · Post by **Killswitch** » Sun Dec 18, 2005 12:28 am

I'm writing an interpreter atm. It's called 'Spectacle', there's some (PureBasic) source floating about on the forums - but it's quite old now. Have a look for it.

I don't think it's that slow!

thefool · Post by **thefool** » Sun Dec 18, 2005 12:30 am

MrMat wrote:NetBeans uses Swing in which each gadget is effectively hand-drawn. This is slower but fits in with Sun's "write once, run anywhere" mantra because it is independent of the OS it is running on. Another graphical interface library is SWT, as used by Eclipse, which acts as a wrapper around the native OS GUI routines, just like PB does as is thus faster and feels more "natural". On the downside this approach requires updating the library for each OS it is to run on. It is up to the author what they use.

Wise words, mr. maD

however it runs quite good here, actually.
5 seconds startup isnt too bad.