I noticed that some pretty important bugs (specifically the ones w.r.t. inout) recently got fixed.  This is a major step forward since it allows containers to be written that work properly with const.  Is it time to start thinking about another release soon?

On Monday, October 03, 2011 00:10:31 David Simcha wrote:
> I noticed that some pretty important bugs (specifically the ones w.r.t. inout) recently got fixed.  This is a major step forward since it allows containers to be written that work properly with const.  Is it time to start thinking about another release soon?

I think that there are a couple of pull requests to druntime/phobos that I'd like to see merged in first, and I'd really like the issue with improvements to deprecated to be sorted out first if at all possible, but on the whole, given some of the important bugs fixed lately, we should probably at least think about doing a release relatively soon.

- Jonathan M Davis

On 10/2/2011 9:10 PM, David Simcha wrote:
> I noticed that some pretty important bugs (specifically the ones w.r.t. inout) recently got fixed.  This is a major step forward since it allows containers to be written that work properly with const.  Is it time to start thinking about another release soon?

Considerable (but apparently incomplete) work has gone into fixing the tangle of bugs surrounding const and ref:

https://github.com/D-Programming-Language/dmd/pull/114 https://github.com/D-Programming-Language/dmd/pull/115

Issue 4251
Issue 5493
Issue 6670 (should this just be closed as a dup of 5493?)

Daniel, you'd been working on it.. on your list of things to finish soon or did you back burner it due to the problems talked about at the end of the comments of pull 115?

---

Additionally, the gc that went in to the win32 build needs to either be disabled or better tuned.  The performance regression it introduces is pretty awful.  If it's going to go in, it really ought to go into all platforms, not just win32.

--

Lastly, there's a bunch of open regressions that are new since the last release.  It'd be nice to get all of those fixed before a release too.  We're back up to 52 regressions+blockers after getting to a low of 48.

My 2 cents,
Brad

On 10/2/2011 9:10 PM, David Simcha wrote:
> I noticed that some pretty important bugs (specifically the ones w.r.t. inout) recently got fixed.  This is a major step forward since it allows containers to be written that work properly with const.  Is it time to start thinking about another release soon?
>

That was a pretty significant change, so I'd like to let that percolate a bit before releasing it.

On Mon, Oct 3, 2011 at 3:35 PM, Brad Roberts <braddr at puremagic.com> wrote:
>
> Considerable (but apparently incomplete) work has gone into fixing the tangle of bugs surrounding const and ref:
>
> https://github.com/D-Programming-Language/dmd/pull/114 https://github.com/D-Programming-Language/dmd/pull/115
>
> Issue 4251
> Issue 5493
> Issue 6670 (should this just be closed as a dup of 5493?)
>
> Daniel, you'd been working on it.. on your list of things to finish soon or did you back burner it due to the problems talked about at the end of the comments of pull 115?
>

4251: The patch I made was wrong, and I now think it's the wrong approach.  I have hopes fixing that will fix the implicit conversion issues found in the const(Object)ref patch, but will likely require non-trivial changes to the template deduction and implicit conversion code.  So, not until the release after next at the earliest.

5493: Same issues as 4251, but can be mostly fixed by pull 114 or similar.  It might be worth reopening 114 just to clear up the most common cases.

6670 is _not_ a dupe of 5493, although the original report seems to be a dupe of 6669.  There are at least two new bugs in that report, invalid optimizing of a variable to its initializer and some incorrect type casting when shared is involved.

Essentially, no complete patches or solutions, and probably no time for the next few weeks.

I'd like to merge two fixes before next release (will be 2.056).

https://github.com/D-Programming-Language/phobos/pull/276
Current unit tests of std.datetime and std.windows.registry are broken
in Windows with multi-byte code pages
(At least Japanese locale).

https://github.com/D-Programming-Language/dmd/pull/433
My inout patch was merged, but it wasn't completion. I think this
patch will fix remains.

Kenji Hara

2011/10/3 David Simcha <dsimcha at gmail.com>:
> I noticed that some pretty important bugs (specifically the ones w.r.t.
> inout) recently got fixed. ?This is a major step forward since it allows
> containers to be written that work properly with const. ?Is it time to start
> thinking about another release soon?
> _______________________________________________
> dmd-internals mailing list
> dmd-internals at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-internals
>

https://github.com/D-Programming-Language/dmd/pull/439 Additional fix for inout range.

Kenji Hara

2011/10/6 kenji hara <k.hara.pg at gmail.com>:
> I'd like to merge two fixes before next release (will be 2.056).
>
> https://github.com/D-Programming-Language/phobos/pull/276
> Current unit tests of std.datetime and std.windows.registry are broken
> in Windows with multi-byte code pages
> (At least Japanese locale).
>
> https://github.com/D-Programming-Language/dmd/pull/433
> My inout patch was merged, but it wasn't completion. I think this
> patch will fix remains.
>
> Kenji Hara
>
> 2011/10/3 David Simcha <dsimcha at gmail.com>:
>> I noticed that some pretty important bugs (specifically the ones w.r.t.
>> inout) recently got fixed. ?This is a major step forward since it allows
>> containers to be written that work properly with const. ?Is it time to start
>> thinking about another release soon?
>> _______________________________________________
>> dmd-internals mailing list
>> dmd-internals at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/dmd-internals
>>
>

FWIW, I'm trying to get dcollections compiled with full inout support.? I think this should be a pretty good test of how usable inout is (it fleshed out this bug that Kenji just fixed).? Should be done this week.


-Steve




>________________________________
>From: kenji hara <k.hara.pg at gmail.com>
>To: Discuss the internals of DMD <dmd-internals at puremagic.com>
>Sent: Saturday, October 8, 2011 7:52 AM
>Subject: Re: [dmd-internals] Time for new release?
>
>https://github.com/D-Programming-Language/dmd/pull/439 Additional fix for inout range.
>
>Kenji Hara
>
>2011/10/6 kenji hara <k.hara.pg at gmail.com>:
>> I'd like to merge two fixes before next release (will be 2.056).
>>
>> https://github.com/D-Programming-Language/phobos/pull/276
>> Current unit tests of std.datetime and std.windows.registry are broken
>> in Windows with multi-byte code pages
>> (At least Japanese locale).
>>
>> https://github.com/D-Programming-Language/dmd/pull/433
>> My inout patch was merged, but it wasn't completion. I think this
>> patch will fix remains.
>>
>> Kenji Hara
>>
>> 2011/10/3 David Simcha <dsimcha at gmail.com>:
>>> I noticed that some pretty important bugs (specifically the ones w.r.t.
>>> inout) recently got fixed. ?This is a major step forward since it allows
>>> containers to be written that work properly with const. ?Is it time to start
>>> thinking about another release soon?
>>> _______________________________________________
>>> dmd-internals mailing list
>>> dmd-internals at puremagic.com
>>> http://lists.puremagic.com/mailman/listinfo/dmd-internals
>>>
>>
>_______________________________________________
>dmd-internals mailing list
>dmd-internals at puremagic.com
>http://lists.puremagic.com/mailman/listinfo/dmd-internals
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/dmd-internals/attachments/20111012/cf7662b9/attachment.html>

Finished, dcollections is now fully inout-aware, and passes all unit tests.? One additional somewhat annoying issue I found:? http://d.puremagic.com/issues/show_bug.cgi?id=6809


I had to deal with some interesting issues (not bugs, but design-related), which I'm going to detail in a newsgroup post.

But one nice thing -- I needed to add 0 overloads to deal with const and immutable versions of collections.? It looks like inout is pretty effective!? In most cases I just had to add the inout decorations.

Here is the related commit:

http://www.dsource.org/projects/dcollections/changeset/114

-Steve




>________________________________
>From: Steve Schveighoffer <schveiguy at yahoo.com>
>To: Discuss the internals of DMD <dmd-internals at puremagic.com>
>Sent: Wednesday, October 12, 2011 3:21 PM
>Subject: Re: [dmd-internals] Time for new release?
>
>
>FWIW, I'm trying to get dcollections compiled with full inout support.? I think this should be a pretty good test of how usable inout is (it fleshed out this bug that Kenji just fixed).? Should be done this week.
>
>
>
>-Steve
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/dmd-internals/attachments/20111013/15d5983f/attachment-0001.html>

On 10/13/2011 5:40 AM, Steve Schveighoffer wrote:
> Finished, dcollections is now fully inout-aware, and passes all unit tests. One additional somewhat annoying issue I found: http://d.puremagic.com/issues/show_bug.cgi?id=6809
>
> I had to deal with some interesting issues (not bugs, but design-related), which I'm going to detail in a newsgroup post.
>
> But one nice thing -- I needed to add 0 overloads to deal with const and immutable versions of collections.  It looks like inout is pretty effective! In most cases I just had to add the inout decorations.
>
> Here is the related commit:
>
> http://www.dsource.org/projects/dcollections/changeset/114
>

This is great news! Thanks to you & Kenji.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/dmd-internals/attachments/20111013/f76e13a7/attachment.html>