Lars Gullik Bjønnes wrote:
> Also to make the process more visible.
Well, that bit worked ;-)
--
Angus
Angus Leeming <[EMAIL PROTECTED]> writes:
| More generally, why submit a patch for inspection if you don't intend
| the final code to look anything like it?
So that you can tell me if it collides with something you work on.
Also to make the process more visible.
--
Lgb
Lars Gullik Bjønnes wrote:
> except for hassle if I want to change the 'Foo' into 'Bar' (or
> course then I'll have other things to fix as well...
> Most likely the whole _fwd file will go, but it is practical for my
> when I work on the dispatch stuff.
Why not say this at the start and avoid rai
On Wed, Oct 29, 2003 at 09:44:46AM +0100, Lars Gullik Bjønnes wrote:
> Andre Poenitz <[EMAIL PROTECTED]> writes:
>
> [IMHOs]
>
> But I don't claim to get sick by seeing your IMHOs.
Well, your stomach seems to be better than mine then.
> And I disagree with several of them, both on subjective re
Andre Poenitz <[EMAIL PROTECTED]> writes:
[IMHOs]
But I don't claim to get sick by seeing your IMHOs.
And I disagree with several of them, both on subjective reason, but
also on technical reasons.
--
Lgb
Andre Poenitz <[EMAIL PROTECTED]> writes:
| On Wed, Oct 29, 2003 at 09:17:40AM +0100, Lars Gullik Bjønnes wrote:
>> Andre Poenitz <[EMAIL PROTECTED]> writes:
>>
>> | Please not. Forward declaring a class be saying 'class Foo;' is standard
>> | practice. It is cheap. And it is as explicit as it ca
On Wed, Oct 29, 2003 at 09:22:57AM +0100, Lars Gullik Bjønnes wrote:
> Andre Poenitz <[EMAIL PROTECTED]> writes:
>
> | On Tue, Oct 28, 2003 at 08:31:04PM +0100, Lars Gullik Bjønnes wrote:
> >> IMHO we should never manually forward declare, that should be the
> >> purpose of a header file instead.
On Wed, Oct 29, 2003 at 09:17:40AM +0100, Lars Gullik Bjønnes wrote:
> Andre Poenitz <[EMAIL PROTECTED]> writes:
>
> | Please not. Forward declaring a class be saying 'class Foo;' is standard
> | practice. It is cheap. And it is as explicit as it can get. Nothing
> | hidden, all clear. #include ca
Andre Poenitz <[EMAIL PROTECTED]> writes:
| On Tue, Oct 28, 2003 at 08:31:04PM +0100, Lars Gullik Bjønnes wrote:
>> IMHO we should never manually forward declare, that should be the
>> purpose of a header file instead.
>
| Could we try to reach an agreement that developers do not install there
| p
Andre Poenitz <[EMAIL PROTECTED]> writes:
| Please not. Forward declaring a class be saying 'class Foo;' is standard
| practice. It is cheap. And it is as explicit as it can get. Nothing
| hidden, all clear. #include can mean everything.
when the include has the postfix "_fwd.h" its intent is pre
On Tue, Oct 28, 2003 at 08:31:04PM +0100, Lars Gullik Bjønnes wrote:
> IMHO we should never manually forward declare, that should be the
> purpose of a header file instead.
Could we try to reach an agreement that developers do not install there
personal pets in the source, especially if it is neit
On Tue, Oct 28, 2003 at 08:31:04PM +0100, Lars Gullik Bjønnes wrote:
> Angus Leeming <[EMAIL PROTECTED]> writes:
>
> | I don't see the point to the _fwd.h file.
> >
> | #include "dispatchresult_fwd.h"
> >
> | is more typing than
> >
> | class DispatchResult;
>
> Well... but now I can add other s
On Tue, Oct 28, 2003 at 07:50:14PM +0100, Lars Gullik Bjønnes wrote:
>
>
> This replaces dispatch_result enum with a real type, forward declared
> when possible.
>
> a couple of Mathed :: dispatch changed to priv_dispatch and a couple
> moved from public to protected.
>
> This will go in unless
Angus Leeming <[EMAIL PROTECTED]> writes:
>> it will be.
>
| I look forward to seeing your solution.
Nothing revolutionary, I asure you.
--
Lgb
Lars Gullik Bjønnes wrote:
>>> This replaces dispatch_result enum with a real type, forward
>>> declared when possible.
>>
> | Also, you don't need to use DispatchResult in the body of the
> | code, just when passing it around. The operator
> | dispatch_result_t() means that things will just work.
Angus Leeming <[EMAIL PROTECTED]> writes:
| I don't see the point to the _fwd.h file.
>
| #include "dispatchresult_fwd.h"
>
| is more typing than
>
| class DispatchResult;
Well... but now I can add other stuff to the header file wihtout
touching a lot of files. (not that many in this case...)
|
Angus Leeming <[EMAIL PROTECTED]> writes:
| Lars Gullik Bjønnes wrote:
>
>>
>> This replaces dispatch_result enum with a real type, forward
>> declared when possible.
>
| Also, you don't need to use DispatchResult in the body of the code,
| just when passing it around. The operator dispatch_resu
Lars Gullik Bjønnes wrote:
>
> This replaces dispatch_result enum with a real type, forward
> declared when possible.
>
> a couple of Mathed :: dispatch changed to priv_dispatch and a couple
> moved from public to protected.
Hey! You're stealing my ideas ;-)
> This will go in unless I get obje
Lars Gullik Bjønnes wrote:
>
> This replaces dispatch_result enum with a real type, forward
> declared when possible.
Also, you don't need to use DispatchResult in the body of the code,
just when passing it around. The operator dispatch_result_t() means
that things will just work.
For example
19 matches
Mail list logo
