[5.10] Native types can't be used with pointers

[IceSoft]

Native types really STILL CAN BE USED with pointers.

since this is probably due to the fact PB doesn't check by design for an existing definition of a structure when one is referenced using pointers inside a structure.

Where is this statement?

[Psychophanta]

@luis, Sorry but you seem to talk about other things.
What i am referring to is other issue. Check the title of this thread please.

[PMV]

Requested for years about this needed fixation for consistence and readability, and finally here it is.

However, there is still the buggy feature:
Native types really STILL CAN BE USED with pointers.
Look & see:

Code: Select all

Structure so
  *locationinmem.b; <- ??
EndStructure
f.so\locationinmem=78

Debug f\locationinmem

I have not installed PB 5.10 here, so can't check if this really works.
But if that is the case as you write, please make a bug-report

MFG PMV

[luis]

Where is this statement?

No idea, probably nowhere. Just try it:

Code: Select all

Structure mela
 item.pera ; not allowed
EndStructure

Code: Select all

Structure mela
 *item.pera ; allowed
EndStructure

@luis, Sorry but you seem to talk about other things.
What i am referring to is other issue.

What are you referring to then ?
Haven't you defined a pointer to a native type (now forbidden) inside a structure ?
If so, I'm talking about the same thing.

Check the title of this thread please.

Ok, wait... Done. It's the same title I was remembering.

You better report it but don't know if Fred will fixed it, since this is probably due to the fact PB doesn't check by design for an existing definition of a structure when one is referenced using pointers inside a structure.

Are we ok now ?

[Psychophanta]

I am full of reporting bugs and lots of them are rejected, so do report you if you want...

[luis]

So we are ok now and we were referring to the same thing, right ?

OK !

[Psychophanta]

So we are ok now and we were referring to the same thing, right ?

OK !

Tha case is that there is allowed to define pointers with native types, that's all.
If the root (cause) of it is that there is allowed to define pointers with any type, including not existing ones, it is other (still worse) thing.

[luis]

If the root (cause) of it is that there is allowed to define pointers with any type, including not existing ones, it is other (still worse) thing.

It's not a bug, it's a special case by design or you couldn't do something like this:

Code: Select all

Structure t1
 *item.t2
EndStructure

Structure t2
 *item.t1   
EndStructure

An item inside a structure of type t2 could point to a structure t1, and an item inside a structure of type t1 could point to a structure t2. But if PB blocked you with an error when you were trying to define structure t1 you would be unable to describe this situation in code.

And yes, this is probably why the ability to define a pointer to a native type slipped through in this case.

Anyway, I'll do the bug report then

[Psychophanta]

I disagree with that "by design feature", because, even with a single compiling pass, the compiler should register (tracking) all the things it is reading, and check whether there is defined later or not...
You see what i mean?

Besides of that, why this doesn't work?. There is another inconsistence, you see?

Code: Select all

Structure t1
  item2.t2
EndStructure

Structure t2
 *item.t1   
EndStructure

[luis]

No, because it's generating code on the way down while it's reading the source.

[Psychophanta]

No, because it's generating code on the way down while it's reading the source.

That is dangerous, and for sure in the future Fred will realize of it.

[luis]

Besides of that, why this doesn't work?. There is another inconsistence, you see?

Code: Select all

Structure t1
  item2.t2
EndStructure

Structure t2
 *item.t1   
EndStructure

No again for the same reason above. t2 is not referenced in your case, it's part of the structure you are trying do define.
In case of the pointer, it's just a pointer, in case of the whole structure the compiler need to know the structure definition.

[Fred]

It's now forbidden as well (to forbid native type in pointer structure field), for consistency.

[Psychophanta]

It's now forbidden as well (to forbid native type in pointer structure field), for consistency.

Still, that kind of issues have a real solution in one pass compiling scheme.
I don't say to do it now, but to take in account for future would be a good thing.

[Psychophanta]

@Fred, at the moment, to give an idea to solve this, i suggest just to leave it as it is now, but just invalidating only the native types (.c , .b , etc.) for the pointers inside structures.