D 1.0 for Jan 1, 2007

[Lars Ivar Igesund]

Don Clugston wrote:

> Lars Ivar Igesund wrote:
>> Brad Roberts wrote:
>> 
>>> Walter Bright wrote:
>>>> Brad Roberts wrote:
>>>>> Can you clarify for the rest of us what you consider this milestone
>>>>> meaning / containing?  What changes at that point in time?  What do
>>>>> you consider finished, frozen, whatever?
>>>>>
>>>>> This has come up each time a 1.0 point has been suggested, with lots
>>>>> of potential things to fall into the bucket, but I don't recall ever
>>>>> seeing your point of view.
>>>>>
>>>>> I'm purposely not including my thoughts or any of the past thoughts
>>>>> from other people since I'm seeking a fairly complete response from
>>>>> you rather than a short agreement sort of answer.
>>>>
>>>> Feature complete is one of those mirage things, the list of what needs
>>>> to be added to D is never ending and ever changing. So my intention is
>>>> to straighten out what we've got now as best I can, call it 1.0, and we
>>>> can all move on doing continuous improvement.
>>> I'm not fishing for new features to be included.  Quite the opposite
>>> actually, considering how little time there is between now and jan 1.
>>> I'll spell out a little more for the type of info I'm looking for,
>>> almost exclusively based on past threads on 1.0 topics:
>>>
>>> 1) is jan 1 a branch point for the language spec?  the compiler?  both?
>>>
>>> 2) if it includes the compiler, are there going to be release
>>> candidates?  before or starting on jan 1?
>>>
>>> 3) if it includes the spec, I assume that you'll (and hopefully the
>>> community) will be spending time doing a thorough review of it both as
>>> it stands and as it is reflected in the compiler.
>>>
>>> 4) if it's the compiler and NOT the spec, what's that mean for the spec?
>>>
>>> 5) if it's the sped and not the compiler, when might the compiler be
>>> frozen to match the spec?
>>>
>>> ... there's bound to be more answerable questions that stem from these
>>> first few.
>>>
>>> Clearer?
>>>
>>> Later,
>>> Brad
>> 
>> I heartily agree with you Brad, the very least I expect from a D 1.0
>> release, is that the spec is frozen/versioned/branched. With this follows
>> that either the compiler is branched (which technically might be the best
>> solution), or allows a way to specify which version of the spec to
>> compile for (preferably latest stable spec as default, with some
>> -experimental switch for new features). Without this, there will be _no_
>> gain whatsoever in proclaiming D 1.0, as users still would have to
>> upgrade to the latest compiler with the newest and greatest just to have
>> bugfixes for the old issues also pertaining to the stable version of the
>> spec.
> 
> You're overstating the case, I think.
> D 1.0 also means:
> * all previous versions are tossed away.
> * all libraries synchronise to the same compiler version. This is very
> important, and has never happened before.
> 
> Instead of 175 possible compiler releases to choose, there will be only
> one obvious choice. Especially for library developers.

I'm not at all worried by all those prior versions, but about DMD 1.01
released 20th of January with some highly needed bugfixes and another great
new (buggy) feature breaking 1.0 code.

-- 
Lars Ivar Igesund
blog at http://larsivi.net
DSource & #D: larsivi