I see that Walter just fixed a typedef bug in 2.056, even though I was just ticked of for even thinking of using one ;=)

Steve

On 03-11-2011 17:22, Steve Teale wrote:
> I see that Walter just fixed a typedef bug in 2.056, even though I was
> just ticked of for even thinking of using one ;=)
>
> Steve

Shouldn't that be long deprecated in favor of alias...?

- Alex

On Thu, 03 Nov 2011 17:28:53 +0100, Alex Rønne Petersen wrote:

> On 03-11-2011 17:22, Steve Teale wrote:
>> I see that Walter just fixed a typedef bug in 2.056, even though I was just ticked of for even thinking of using one ;=)
> 
> Shouldn't that be long deprecated in favor of alias...?

That's what everybody told me. Since it is not perfect it must die.

Steve

On Thursday, November 03, 2011 09:22 Steve Teale wrote:
> I see that Walter just fixed a typedef bug in 2.056, even though I was just ticked of for even thinking of using one ;=)

He probably fixed it because it hasn't actually been given the axe yet, but it's definitely going to get the axe - and probably soon, since there was some discussion of wanting to remove (or at least deprecate) features that are definitely going to be removed before gdc actually makes it into gcc so that the changes are less disruptive when they happen.

- Jonathan M Davis

On Thu, 03 Nov 2011 13:22:49 -0400, Jonathan M Davis wrote:

> On Thursday, November 03, 2011 09:22 Steve Teale wrote:
>> I see that Walter just fixed a typedef bug in 2.056, even though I was just ticked of for even thinking of using one ;=)
> 
> He probably fixed it because it hasn't actually been given the axe yet, but it's definitely going to get the axe - and probably soon, since there was some discussion of wanting to remove (or at least deprecate) features that are definitely going to be removed before gdc actually makes it into gcc so that the changes are less disruptive when they happen.
> 
Jonathan,

I thank you for your reply. I was just watching Sarkozy at the G20 talking about the potential Greek referendum. He took a similarly firm view. The outcome of that situation remains TBD.

I wonder if D in gcc might be rather like the Eurozone.

Steve

On Thu, 03 Nov 2011 18:22:49 +0100, Jonathan M Davis <jmdavisProg@gmx.com> wrote:

> On Thursday, November 03, 2011 09:22 Steve Teale wrote:
>> I see that Walter just fixed a typedef bug in 2.056, even though I was
>> just ticked of for even thinking of using one ;=)
>
> He probably fixed it because it hasn't actually been given the axe yet, but
> it's definitely going to get the axe - and probably soon, since there was some
> discussion of wanting to remove (or at least deprecate) features that are
> definitely going to be removed before gdc actually makes it into gcc so that
> the changes are less disruptive when they happen.
>
> - Jonathan M Davis

I think we need to change the release policy to make that happen.
Honestly the half-life period for D code using the mainline dmd
is about a few weeks and is released on ~monthly base.
This is a heavy restriction to attract professional development.

Currently this is creates the unlucky situation of deferring useful changes
to some projected D3, while still breaking the D language all too often.

I think there's a lot to learn from http://python.org/download/releases/.
For example somebody changing from Python2.5 to 2.7 is anticipating
some breaking changes, not so much for a change from dmd2.053 to dmd2.056.

martin

On Thu, 3 Nov 2011, Martin Nowak wrote:

> I think we need to change the release policy to make that happen.
> Honestly the half-life period for D code using the mainline dmd
> is about a few weeks and is released on ~monthly base.
> This is a heavy restriction to attract professional development.
> 
> Currently this is creates the unlucky situation of deferring useful changes to some projected D3, while still breaking the D language all too often.
> 
> I think there's a lot to learn from http://python.org/download/releases/.
> For example somebody changing from Python2.5 to 2.7 is anticipating
> some breaking changes, not so much for a change from dmd2.053 to dmd2.056.
> 
> martin

While I don't disagree with the larger point, I do feel compelled to clarify that it's not dmd or the language that's making most of the backwards incompatible changes, it's phobos.

Walter has been considerably more conservative in what breaking changes are acceptable.

On Thu, 03 Nov 2011 19:44:36 +0100, Brad Roberts <braddr@puremagic.com> wrote:

> On Thu, 3 Nov 2011, Martin Nowak wrote:
>
>> I think we need to change the release policy to make that happen.
>> Honestly the half-life period for D code using the mainline dmd
>> is about a few weeks and is released on ~monthly base.
>> This is a heavy restriction to attract professional development.
>>
>> Currently this is creates the unlucky situation of deferring useful changes
>> to some projected D3, while still breaking the D language all too often.
>>
>> I think there's a lot to learn from http://python.org/download/releases/.
>> For example somebody changing from Python2.5 to 2.7 is anticipating
>> some breaking changes, not so much for a change from dmd2.053 to dmd2.056.
>>
>> martin
>
> While I don't disagree with the larger point, I do feel compelled to
> clarify that it's not dmd or the language that's making most of the
> backwards incompatible changes, it's phobos.
>
> Walter has been considerably more conservative in what breaking changes
> are acceptable.

True, but that has it's own drawbacks.
Also adding bugs for new features can be considered as a breaking change.
There's a lot to gain from splitting maintenance and feature development.

martin

Am 03.11.2011, 18:43 Uhr, schrieb Steve Teale <steve.teale@britseyeview.com>:

> On Thu, 03 Nov 2011 13:22:49 -0400, Jonathan M Davis wrote:
>
>> On Thursday, November 03, 2011 09:22 Steve Teale wrote:
>>> I see that Walter just fixed a typedef bug in 2.056, even though I was
>>> just ticked of for even thinking of using one ;=)
>>
>> He probably fixed it because it hasn't actually been given the axe yet,
>> but it's definitely going to get the axe - and probably soon, since
>> there was some discussion of wanting to remove (or at least deprecate)
>> features that are definitely going to be removed before gdc actually
>> makes it into gcc so that the changes are less disruptive when they
>> happen.
>>
> Jonathan,
>
> I thank you for your reply. I was just watching Sarkozy at the G20
> talking about the potential Greek referendum. He took a similarly firm
> view. The outcome of that situation remains TBD.
>
> I wonder if D in gcc might be rather like the Eurozone.
>
> Steve

It caused some discussion. Someone didn't like the naked-function feature or other traits of D that (official) support has to be written for in the GCC backend first. But I don't see how D would become unbearable for the GCC people as long as the frontend is maintained. There have been other language frontends bitrotting before, so the 'Eurozone' is suspicious and looks closely on who is joining them.
Since D in GCC is often requested I believe it will have a positive future there and give the language more momentum, since people don't have to leave their known and pre-installed tool chain.

On 11/3/11 7:44 PM, Brad Roberts wrote:
> While I don't disagree with the larger point, I do feel compelled to
> clarify that it's not dmd or the language that's making most of the
> backwards incompatible changes, it's phobos.

I don't think this is generally true – for template metaprogramming-heavy code, DMD changes (regressions/bugfixes subtly altering behavior of some of the less-defined areas of the language) usually cause me much more of a headache than Phobos does.

David