Pure FB Runtime Library (in progress)

[MrSwiss]

> ... (all common knowledge)

Why, for Pete's sake, can't you ever admit, to have made a mistake?
Would probably make you look, a bit more human?

[St_W]

Code: Select all

VarPtr(array(0))  ' or
VarPtr(LBound(array))

doesn't (and shouldn't) compile:

C:\test.bas(4) error 24: Invalid data types, for @ or VARPTR in 'print varptr(lbound(array))'

//edit: full test code:

Code: Select all

dim array(10) as integer

print @array(0)
print varptr(lbound(array))

sleep

Why, for Pete's sake, can't you ever admit, to have made a mistake?
Would probably make you look a bit more human?

Ask that yourself :-P (before blaming others)

[MrSwiss]

Ask that yourself :-P

It works very well, in a real example: (no errors!)

Code: Select all

Dim As ULong colour(0 To ...) = { &hFF000000, &hFF3F3F3F, &hFF7F7F7F }

Print VarPtr(colour(0))

Sleep

[fxm]

VarPtr used as follows, on a array:

Code: Select all

VarPtr(array(0))  ' or
VarPtr(LBound(array))

The syntax problem is on the second line.

[MrSwiss]

The syntax problem is on the second line.

OK, thanks, duly noted ...
The first one, however, which proves the point, works ...

[fxm]

That proves nothing at all because "array (0)" is a particular element of array, and not the array itself.

When we have a string (s), we can use not only:
VarPtr(s[0]) '' address of the first element of the character string
but also:
StrPtr(s) '' start address of the character string

[MrSwiss]

Are you aware of the fact, that for practical use, both of them are identical?
VarPtr(s[0]) = VarPtr(s) -- the very same Address ...
A String is (sort of, apart from "header"), a array of UByte (ASCII-String only).

Only: VarPtr(s[n]) would be different ... (where: n > 0)

[fxm]

Yes, these two entities have the same value, but they are entities of different types:

Code: Select all

Dim As String s

#print typeof(VarPtr(s[0]))
#print typeof(Strptr(s))

UBYTE PTR
ZSTRING * 4 CONST PTR

Fortunately, there is compatibility between "Ubyte Ptr" and "Zstring Ptr" for assignment, but not fully compatibility between "Ubyte" and "Zstring":

Code: Select all

Dim As String s = "FreeBASIC"

#print typeof(VarPtr(s[0]))
#print typeof(Strptr(s))
Dim As Ubyte Ptr pu = Strptr(s)       '' OK
Dim As Zstring Ptr pz = VarPtr(s[0])  '' OK

#print typeof(*VarPtr(s[0]))
#print typeof(*Strptr(s))
Dim As Ubyte u = *Strptr(s)           '' OK
Dim As Zstring *10 z = *VarPtr(s[0])  '' error 24: Invalid data types

UBYTE PTR
ZSTRING * 4 CONST PTR
UBYTE
ZSTRING
.....\FBIde0.4.6r4_fbc1.06.0\FBIDETEMP.bas(11) error 24: Invalid data types.....

[MrSwiss]

Now, that has nothing any longer to do, with the initial question, of:
"How to get the *start*-address of an array, with a FB-keyword?" since you stated:
@ is a *short-cut* of VarPtr ... (@ = symbol (for: "get address"), VarPtr = keyword)

[Imortis]

I have a git repo for this project here: https://github.com/ImortisInglorian/fbrtLib
At this point I have only confirmed that it will compile with what I have. I have been going back and fixing bugs in my translations in older files as they crop up during the compile process. I am using posidenFB as my IDE and the project file for it is in the repo. I am compiling it as a static library as I figured that would make testing it much easier when I have something testable.

It is not in a usable state.

If you want to eye-ball the code and point out how stupid I am or how many errors I have made (please be specific), feel free. I have commented out things that are not yet needed, and left thing in that will not compile on a windows environment since that is what I am working in.

I am getting 5 to 6 files converted a day, which is not great, but it's what I can do and still get work done and have a life outside of work and coding.

[counting_pine]

[abridged]

Code: Select all

static FB_STR_TMPDESC fb_tmpdsTB[FB_STR_TMPDESCRIPTORS];

I assumed the first was just an array declare, roughly this in FB:

Code: Select all

dim fb_tmpdsTB(FB_STR_TMPDESCRIPTORS) as FB_STR_TMPDESC

Declaring an array of size N in C gives you N elements starting at 0, i.e. 0 through N-1.
So it's necessary to adjust the array subscripts:

Code: Select all

dim shared fb_tmpdsTB(0 to FB_STR_TMPDESCRIPTORS-1) as FB_STR_TMPDESC

(I can see you've already added the 'shared' in https://github.com/ImortisInglorian/fbr ... re.bas#L20)

'dim array(0 to N-1)' is/should be pretty much considered as an idiom in FB, since 'dim a(N)' doesn't do the same thing, and 'dim a(N-1)' looks weird.

[Imortis]

[abridged]

Code: Select all

static FB_STR_TMPDESC fb_tmpdsTB[FB_STR_TMPDESCRIPTORS];

I assumed the first was just an array declare, roughly this in FB:

Code: Select all

dim fb_tmpdsTB(FB_STR_TMPDESCRIPTORS) as FB_STR_TMPDESC

Declaring an array of size N in C gives you N elements starting at 0, i.e. 0 through N-1.
So it's necessary to adjust the array subscripts:

Code: Select all

dim shared fb_tmpdsTB(0 to FB_STR_TMPDESCRIPTORS-1) as FB_STR_TMPDESC

(I can see you've already added the 'shared' in https://github.com/ImortisInglorian/fbr ... re.bas#L20)

'dim array(0 to N-1)' is/should be pretty much considered as an idiom in FB, since 'dim a(N)' doesn't do the same thing, and 'dim a(N-1)' looks weird.

Thanks counting_pine. Already fixed in my local version.

[Imortis]

Okay, next question:

Code: Select all

LenAdd = sprintf( ((pszAdd = FixPart), FixPart)"%d", fb_Day( value ) );

How is that valid? FixPart is a char array and spzAdd is a char pointer.

EDIT: Whoops, never mind. I accidentally removed a comma from the line of code.

EDIT #2: By the way: the code behind the "FORMAT" command is a real mess. There has got to be a better way to do it. Maybe after this project is finished I, or someone else, can take a stab at rewriting it.

[counting_pine]

https://github.com/freebasic/fbc/blob/m ... mat.c#L843?
Potentially when writing in FB, it might be possible to simplify some code by using strings and string functions.
It might not be as fast or memory-efficient. But on the other hand, simpler implementations might be a better foundation for writing/testing more efficient code based on it.
I'd say, feel fee to make the code nicer/simpler if you want. It's one of the main advantages to be gained from writing in FB. It will also help the understanding of the code.

[TeeEmCee]

Wow, you've done a lot of work already! Have you tried running it at all?

I notice a mistake in the translation: you've been translating 'int' to 'integer', but the equivalent of C's 'int' in FB is 'long'.
Also, you need to wrap everything in Extern "C" if you're not doing that already.