Initial announcement : http://forum.dlang.org/post/ld2586$17f6$1@digitalmars.com
Documentation        : http://www.walterbright.com/tmp/html/scopebuffer.html
Code                 : https://github.com/D-Programming-Language/phobos/pull/1911

Starting formal review thread for Walter's proposal.

Linked pull request already has some ongoing discussion and is worth investigating. In this thread please focus on those high-level properties of reviewed module:

 - Is it useful for standard library?
 - Does it fit into Phobos design?
 - Is documentation / API clear enough for you?
 - Will it be possible to keep it within its current API without breaking changes?
 - Are there any major oversights in existing implementation?

Smaller style / implementation comments are also welcome but less important at current stage.

Review will last for 2 weeks (ending on March 30) but can be prolonged upon explicit request.

On Sunday, 16 March 2014 at 19:53:53 UTC, Dicebot wrote:
> Initial announcement : http://forum.dlang.org/post/ld2586$17f6$1@digitalmars.com
> Documentation        : http://www.walterbright.com/tmp/html/scopebuffer.html
> Code                 : https://github.com/D-Programming-Language/phobos/pull/1911
>
> Starting formal review thread for Walter's proposal.

Seems like the plan is to see how it works out for druntime/Phobos first: https://github.com/D-Programming-Language/druntime/pull/739.

David

On Sunday, 16 March 2014 at 19:53:53 UTC, Dicebot wrote:
> Review will last for 2 weeks (ending on March 30) but can be prolonged upon explicit request.

Actually I have just remembered that I won't have regular internet access on week after the next one so lets prolong it to April the 2nd :)

On Sunday, 16 March 2014 at 20:08:32 UTC, David Nadlinger wrote:
> Seems like the plan is to see how it works out for druntime/Phobos first: https://github.com/D-Programming-Language/druntime/pull/739.
>
> David

I think that discussion happening there is much better suited to NG review thread, it is very important topic that should have full community exposure. Even if it does not fit into normal Phobos review limits (breaking the rules, yay!)

I've started using it in a project to replace a simple dynamic array buffer and so far it has fit my needs very well. One thing I ran into was a method in the project's API that has this signature:

const(ubyte[]) foo() const

I couldn't return a slice from ScopeBuffer directly because of the const. I resolved it by changing opSlice's signature in ScopeBuffer to this:

@system inout(T[]) opSlice() inout

So there are two questions:

1.) Is this a good way to resolve the problem or am I doing something horribly wrong?
2.) Assuming this isn't horribly wrong should ScopeBuffer use inout like this?

16-Mar-2014 23:53, Dicebot пишет:
> Initial announcement :
> http://forum.dlang.org/post/ld2586$17f6$1@digitalmars.com
> Documentation        :
> http://www.walterbright.com/tmp/html/scopebuffer.html
> Code                 :
> https://github.com/D-Programming-Language/phobos/pull/1911
>
> Starting formal review thread for Walter's proposal.

Seriously, there is no need for that now, since it was re-targeted for internal use only.


-- 
Dmitry Olshansky

On 3/16/14, 3:47 PM, David Eagen wrote:
> I've started using it in a project to replace a simple dynamic array
> buffer and so far it has fit my needs very well. One thing I ran into
> was a method in the project's API that has this signature:
>
> const(ubyte[]) foo() const
>
> I couldn't return a slice from ScopeBuffer directly because of the
> const. I resolved it by changing opSlice's signature in ScopeBuffer to
> this:
>
> @system inout(T[]) opSlice() inout
>
> So there are two questions:
>
> 1.) Is this a good way to resolve the problem or am I doing something
> horribly wrong?
> 2.) Assuming this isn't horribly wrong should ScopeBuffer use inout like
> this?

good way, it should -- Andrei

Closing this seeing as Andrei has just merged it as std.internal into master. I am very angry about the way it has happened.

On Monday, 17 March 2014 at 13:57:59 UTC, Dicebot wrote:
> Closing this seeing as Andrei has just merged it as std.internal into master. I am very angry about the way it has happened.

The fix for pointers hasn't been integrated. "ScopeBuff!(int*)" still doesn't compile.

On 3/17/14, 6:57 AM, Dicebot wrote:
> Closing this seeing as Andrei has just merged it as std.internal into
> master. I am very angry about the way it has happened.

What happened now??

Andrei