Support a 'new' Syntax (Structure pointer as return value)

[IceSoft]

@Fred
Please implemet this new syntax for using of: structure pointer as return value:
(Currently PB returns a syntax error using this new 'keywords')

Code: Select all

Declare.*<Struct> foo()
DeclareC.*<Struct> foo()
DeclareCDLL.*<Struct> foo()
DeclareDLL.*<Struct> foo()

Procedure.*<Struct> foo()
ProcedureC.*<Struct> foo()
ProcedureCDLL.*<Struct> foo()
ProcedureDLL.*<Struct> foo()
    
ImportC
  foo.*<Struct>() As "foo@16"
EndImport
  
Import
  foo.*<Struct>() As "foo@16"
EndImport 

Here an example how it can be used:

Code: Select all

Structure cpVect
  x.d
  y.d
EndStructure

Declare.*cpVect foo()

[do something]

Procedure.*cpVect foo()
  *a.cpVect = AllocateMemory(SizeOf(cpVect))
  *a\x = 11.1
  *a\y = 22.1
  ProcedureReturn *a
EndProcedure  
  

[Danilo]

I use this currently:

Code: Select all

Structure Vect
  x.d
  y.d
EndStructure
Macro pVect : i : EndMacro

Declare.pVect foo()

[do something]

Procedure.pVect foo()
  *a.Vect = AllocateMemory(SizeOf(Vect))
  *a\x = 11.1
  *a\y = 22.1
  ProcedureReturn *a
EndProcedure

I like to add something to the wishlist here, because it is similar:
Please allow use of Interfaces in Declare, Prototype and Procedure return type declarations.

Code: Select all

Interface IDirectX14
  create()
EndInterface

Procedure.IDirectX14 Blah()
  ; blub
EndProcedure

It is just a pointer, so it shouldn't be a problem for Interfaces.

Currently we get the following error message: "A structure can't be used with ProcedureReturn.", which does not make sense at all.
There is no structure used in the code, and there is no ProcedureReturn!

So, allowing Interfaces as return type should be easy, as it's a pointer only anyway...?

For returning real Pointers, IceSoft's example doesn't look so bad to me. Common PureBasic Syntax, everybody will understand it.

[IceSoft]

For returning real Pointers, IceSoft's example doesn't look so bad to me. Common PureBasic Syntax, everybody will understand it.

Thanks!
And it can be very easy internal replaced with ".i" before compiler and syntax check is running over the source file.

Here the small regular expression used internal which improve the PureBasic syntax:

Code: Select all

If CreateRegularExpression(0, "\.\*.*?[ (]")
  a.s = ReplaceRegularExpression(0, source, ".i")
  MessageRequester("Info", "Structure pointer syntax replaced here: "+a)
EndIf

[freak]

http://www.purebasic.fr/english/viewtop ... =3&t=44120
http://www.purebasic.fr/english/viewtop ... =3&t=52428

Please stop asking for the same thing over and over. It is getting annoying.

[IceSoft]

http://www.purebasic.fr/english/viewtop ... =3&t=44120
http://www.purebasic.fr/english/viewtop ... =3&t=52428

Please stop asking for the same thing over and over. It is getting annoying.

An answer like this is also annoying. So stop it please too. Thanks.

[Fred]

It is a common rule to not open several times the same topic on a forum, or it will get messy very quickly.

[freak]

http://www.purebasic.fr/english/viewtop ... =3&t=44120
http://www.purebasic.fr/english/viewtop ... =3&t=52428

Please stop asking for the same thing over and over. It is getting annoying.

An answer like this is also annoying. So stop it please too. Thanks.

Seriously, do you think your chances of getting things implemented are better when you annoy the hell out of the people you are asking? I don't think so.

[IceSoft]

http://www.purebasic.fr/english/viewtop ... =3&t=44120
http://www.purebasic.fr/english/viewtop ... =3&t=52428

Please stop asking for the same thing over and over. It is getting annoying.

An answer like this is also annoying. So stop it please too. Thanks.

Seriously, do you think your chances of getting things implemented are better when you annoy the hell out of the people you are asking? I don't think so.

I am not sure. But anyhow I think you know about this whish and how easy the implementation is
Ok I know YOU are the master and I am only one of your smallest slave of course

[RichAlgeni]

This reply isn't to tell people what to do, and it's not criticize anyone's coding practices. It's just that my experience may be helpful to some.

All my procedures return an integer, positive for successful, 0 for unsuccessful, negative for an error.

Any data that is needed in both the parent and daughter procedure is instantiated (created, if you will), in the parent process. Strings are allocated, number are initialized, etc.

Never pass strings to a procedure, always pass pointers. If a structure needs to be processed in some way, initialize the structure in the calling procedure, pass the pointer, then check to see if the returned variable is a positive integer.

If an unknown amount of data will be sent back to a parent procedure, set a variable with the maximum length in the parent, allocate memory to this length, plus one byte for a terminator, then pass this length and the pointer to the procedure.

Always, always use type '.i' for integers, even if you need just one digit, it will save you in the long run.

Like I said, this is just the (maybe) benefit of my experience, but by using this practices, I have just about eliminated all IMA errors in my code, my procedures are standardized, (see functional coding), and I never have to worry about returning structures or pointers back to a calling process.