[fpc-devel] Wrong docs: not initialized global variables

[Ondrej Pokorny]

On 24.03.2018 23:00, Sven Barth via fpc-devel wrote:
> Ondrej Pokorny <lazarus at kluug.net <mailto:lazarus at kluug.net>> schrieb 
> am Sa., 24. März 2018, 20:49:
>
>     This is not correct. Global simple variables are always
>     initialized. At
>     least in Delphi it is so:
>     http://docwiki.embarcadero.com/RADStudio/Tokyo/en/Variables_(Delphi)
>     <http://docwiki.embarcadero.com/RADStudio/Tokyo/en/Variables_%28Delphi%29>
>     "If
>     you do not explicitly initialize a global variable, the compiler
>     initializes it to 0."
>
>
> So, Delphi *does* document it. Okay, in that case the documentation 
> needs to be updated, cause up to now the assumption has been that this 
> is an implementation detail.

It has always been so in Delphi - at least since I learnt it. The Delphi 
7 docs (I can't find older docs online) state it as well: 
http://docs.embarcadero.com/products/rad_studio/cbuilder6/EN/CB6_ObjPascalLangGuide_EN.pdf 
on page 5-38 "If you don’t explicitly initialize a global variable, the 
compiler initializes it to 0." (Docs are at 
http://docs.embarcadero.com/products/rad_studio/ )

>     Another question about this sentence in FPC docs: "Managed types are
>     always initialized: in general this means setting the reference
>     count to
>     zero, or setting the pointer value of the type to Nil."
>
>     Does it mean I can assume a local string variable is always (=whenever
>     the routine is called) initialized to ''? I.e. that TestB always
>     returns
>     'a'?
>
>
> For managed types this is indeed guaranteed. Otherwise there'd be 
> exceptions upon the first use of the variable as the RTL would assume 
> that the value is valid.

OK, thanks. My assumption was that setting local managed types to nil is 
guaranteed only once at the first routine call. And that it is an 
implementation detail if it is niled at the second call or reused from 
the first call.

I probably mixed it up with the Result variable (what Maciej wrote about).

Ondrej
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freepascal.org/pipermail/fpc-devel/attachments/20180324/20bb7440/attachment.html>