Removed: ASCII mode -- What dooes that mean??

[Randy Walker]

25th July 2016 : Version 5.50
- Removed: ASCII mode for internal PureBasic string representation, PureBasic is now unicode only.

If PureBasic is now unicode only, does that mean *ALL* string funcrions are affected?
Trying to figure out what broke my code so it does not work on the 5.63 compiler and beyond.
I don't do unicode so I don't understand what it is, or how to adapt to this change.
Please pardon my ignorance

[idle]

yes all strings are UCS2 encoded 2 bytes wide

you also have two pseudo types for use with imported function that only accept ascii or utf8 parameters p-ascii and p-utf8
and also two functions ascii(string) utf8(string) which return pointers you need to free them

It doesn't take long to adapt but yes it was a bit of an inconvenience at the time.

and also with latest 6.11 you need to check that your import parameters on x64 are right, a few got caught out using .l when we should have been using .i

[infratec]

We need to know waht you are doing at the location of the fault.
Else we can not provide a solution.

[Randy Walker]

Code: Select all

a$ ="B" ;single Byte
Debug a$ ; returns single byte
Debug Len(a$) ;confirms a$ is one Byte

So I'm confused. Where's the other Byte?

[infratec]

Code: Select all

a$ ="B" ;single Byte -> WRONG -> 2 bytes
Debug a$ ; returns single byte -> WRONG -> returns 1 character
Debug Len(a$) ;confirms a$ is one Byte -> WRONG -> confirms 1 character
Debug StringByteLength(a$)  ; shows the truth
ShowMemoryViewer(@a$, StringByteLength(a$))

[Randy Walker]

oK infratec and thanks for clarifying. I guess I need to look in that direction. I was brought up on the understanding that a character = one Byte so I'm sure everything in my code is based on that assumption.

[Mindphazer]

Well, no matter if you are in ASCII mode or in Unicode mode, Len(a$) will always return 1 in your example.
So if you only work with a$, you should not see any difference

None of my applications ever had problems when PB turned from ASCII to Unicode...

[AZJIO]

Randy Walker
One character is now 2 bytes, no more difference. ASCII has many limitations; Linux uses UTF-8 by default.

[Fred]

Its an under-the-hood change, it shouldn't impact much applications unless you are dealing with external API which requiers ASCII/UTF8 inputs (ie: not PB commands).

[skywalk]

Yes, it was fairly painless.

Code: Select all

Verify your source code:
  Encode all source text files to UTF-8. (I use Notepad++ to confirm)
  Use SizeOf(Character), not +1 for string terminations.
  Change api prototypes exporting Ascii strings:
  FROM: Prototype.i SendAscii(Dest.i, buf.s, nChars.i, *nCharsReturned_i)
  TO:   Prototype.i SendAscii(Dest.i, buf.p-Ascii, nChars.i, *nCharsReturned_i)

[Axolotl]

If you only use the PB String Procedures, everything should be fine.
However, you should check older implementations that work with string pointers.
Because It can also happen within procedures that evaluate strings via pointers and do not use the Character type for this.
Just in case.

[Randy Walker]

yes all strings are UCS2 encoded 2 bytes wide

you also have two pseudo types for use with imported function that only accept ascii or utf8 parameters p-ascii and p-utf8
and also two functions ascii(string) utf8(string) which return pointers you need to free them

It doesn't take long to adapt but yes it was a bit of an inconvenience at the time.

----------------------------------------------------------------------------------------------------------------------------------
This change has prevented me from keeping up with the newer releases. I'm stuck in v5.40 as it is now. Looking to adapt I thought maybe I could code around these Unicode issues if I checked the compiler version but that seems to be broken in v6.21. If I run the following code on my Window 11 Pro machine using v6.21 it tells me I am running v5.40:

Code: Select all

Debug #PB_Compiler_Version
Debug #PB_Compiler_Version

I discovered why my program will not work when compiled under v6.21 when I accidentally set my 5.40 compiler to use Unicode and broke some features in my program. Then I went to try my code with that setting disabled in v6.21c but the setting was gone. I the searched history and found that same snippet posted in my OP:

As of 25th July 2016 : Version 5.50
- Removed: ASCII mode for internal PureBasic string representation, PureBasic is now Unicode only.

Consequent to the above I went and downloaded newer releases.
I got v5.46 to replace my 5.40.
And I got v6.30 to replace my 6.21.
Just for the hell of it, knowing that wont necessarily improve my situation -- Just creates more work for me.

[idle]

There are possibly dozens of reasons a 5.40 program won't compile with 6.x not just the missing ascii functions.
if you use pointers to strings in a loop you should increment by + SizeOf(Character) so it will be backward compatible
If you import structures or functions double check the parameters long vs integer
and you can always ask for help but without code it's hard to help.

[Randy Walker]

There are possibly dozens of reasons a 5.40 program won't compile with 6.x not just the missing ascii functions.

Oh it compiles, no errors doing so. The resulting application is broken though.

[Randy Walker]

if you use pointers to strings in a loop you should increment by + SizeOf(Character) so it will be backward compatible

I'm really going to have to think hard on that one. Not sure how that might apply to my code.