Thread

Thread Index

Message

From: Dieter Baron <dillo%danbala.tuwien.ac.at@localhost>
To: Pierre Joye <pierre.php%gmail.com@localhost>
Subject: Re: 0.10rc1 prerelease
Date: Fri, 18 Feb 2011 14:55:40 +0100

On Feb 18, 2011, at 14:28, Pierre Joye wrote:

> On Fri, Feb 18, 2011 at 2:17 PM, Dieter Baron
> <dillo%danbala.tuwien.ac.at@localhost> wrote:
>> On Feb 18, 2011, at 12:58, Pierre Joye wrote:
> 
>> please use our time on more important issues.
> 
> Sorry to waste my time testing your code, giving feedback, sometimes
> bug fixes and as a side task, trying to make this project move
> forward.

  Your feedback is very much appreciated, sorry if my mail came across 
otherwise.

  However, we had this discussion less than two months ago, and I was annoyed 
by it then.  But since this is obviously a matter of concern for some of you, 
let's discuss it in further detail and settle it once and for all.

  I have a technical reason to do it this way, which I somehow don't seem to be 
able to get across:  uint64_t et. al. are defined by the C99 standard; they are 
reserved types.  If libzip defined them, and another library defined them as 
well, the application including both header files would get a compile error: 
redefinition of typedef uint64_t (yes, this happens even if the two typedefs 
are exactly the same, I just tested it with gcc 4.2.1).

  zip_uint64_t is an attempt to provide a portable, fixed width type.  
Applications can use it interchangably with other types of the same 
width/signedness, without explicit conversion.  So if your application has 
access to fixed witdh types (provided by the compilation environment or defined 
by the application itself), it need not use zip_uint64_t in its code; it can 
use libzip as if the functions were defined using uint64_t.

  So what is your problem with these types, besides that they don't look nice 
(which is the price you often pay for portability, and which need not spread 
beyond libzip's code and header file)?

yours,
dillo

Made by MHonArc.