[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-devel
Subject:    Re: [PATCH] (Was: ksize_t weirdness ..)
From:       Oswald Buddenhagen <ob6 () inf ! tu-dresden ! de>
Date:       2001-04-07 15:34:36
[Download RAW message or body]

> > NOTE: kdecore/ksocks.cpp will probably break on systems where
> > sizeof(ksocklen_t) != sizeof(socklen_t), as it uses a lot of
> > ksocklen_t's where socklen_t is expected. however, i'm not sure
> > which way round to fix it. george?
> 
> I cannot speak for George, but I think that system calls should be cast 
> to socklen_t and that whatever we defined would be something consistent.
> 
yep, the problem is this one:
there is a huge function dispatcher, which calls an appropriate socks
function if available and otherwise the usual socket funct. the question
is, whether one should make the whole dispatcher use socklen_t natively
or make the explicit conversions immediately before calling the
native socket functions. 
however, i didn't look at the code too close, as i'm busy enough with
my favourite toy kdm. :)

> > GENERAL NOTE: the right way (tm) to include config.h (i think :):
> > - it should be quoted with <>, not ""
> 
> Duly noted, but, I thought <> should be used for system headers, "" for 
> local headers, even though most of the time they have the same meaning. 
> 
the exact meaning is this:
<> -> search only the -I directories (including builtin paths)
"" -> search current directory first, then same as <>
at least i think i remember having read something like that once upon
a time. :)
as in kde config.h is never in ., but in the -I paths, <> theoretically
saves a few milliseconds. :)

best regards

-- 
Hi! I'm a .signature virus! Copy me into your ~/.signature, please!
--
Nothing is fool-proof to a sufficiently talented fool.
 
>> Visit http://master.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<

[prev in list] [next in list] [prev in thread] [next in thread]