PB 6.40 a3 Concat with string pointer

[fryquez]

Code: Select all

buf = AllocateMemory(512)
PokeS(buf, "Hallo")

*s.string = @buf
Debug *s\s + "!"

[mk-soft]

No bug

The type string has changed. It is no longer a pointer on a string, but the type string is now a structured type with lengths. It must be a valid PB string.
You must take with PeekS on your memory area.

Code: Select all

text.String\s = "Hello"
*s.string = @text
ShowMemoryViewer(@*s\s - SizeOf(Integer), 32)
Debug *s\s + "!"

[fryquez]

Of course it's a bug. You have the same situation with fixed strings, these also don't have the size prefixed.

For fixed strings PB now makes a temporary copy that than get passed to SYS_ConcatString.
At least the same should be done here.

[ChrisR]

.

[mk-soft]

@Chris,

You write the length of the string outside the allocated memory.

Corrected ...

Code: Select all

buf = AllocateMemory(512)
Lenght = PokeS(buf+SizeOf(Integer), "Hallo")
ShowMemoryViewer(buf, Lenght + SizeOf(Integer))
PokeI(buf,  Lenght / SizeOf(Character))
ShowMemoryViewer(buf, Lenght + SizeOf(Integer))
*str = buf + SizeOf(Integer)
*s.string = @*str
Debug *s\s + "!"

Not Official. You should not use an intera from Purebasic here, because something can always change here. Better to use PeekS!

[ChrisR]

.

[mk-soft]

@Chris

You write your string length again before the allocated memory. Your memory starts at "buf" and not 8 bytes before

[ChrisR]

.

[mk-soft]

Turn on the purifier

You set the pointer *p to 8 bytes before allocated memory and then pokes into the memory before allocated buffer.

[kenmo]

Not a bug... incorrect use of pointer... A string in the new 6.40 system now consists of the character data (with null terminator still), and length info and flag bits...
You can't point a PB 6.40 string pointer directly at the raw character data missing the other info and expect it to work correctly. I think Fred and Freak will confirm this.

EDIT: And ChrisR, you can't allocate a buffer at address (buf) and then write numeric data before (buf) Crashes and bugs will happen!

[ChrisR]

Sorry, I'm preoccupied with other, more important things right now and I shouldn't have written here in the first place. The best I can do is delete my buggy posts

[mk-soft]

Should not delete the bug code. I didn't do it with my bugs and just wrote on how not to do it.