phobos commit, revision 1564


user: walter

msg:
bugzilla 4056 Template instantiation with bare parameter not documented

http://www.dsource.org/projects/phobos/changeset/1564

Excellent! That's #1 from my list of the worst bugs.

There are just two patches I'd like to see in the next release:
* Brad's patch for arguably the worst wrong-code bug:
2127 inliner turns struct "return *this" from by-value into by-ref

*And this simple compiler fault found by a newbie, which is a very
recent regression:
4242 ICE(module.c): importing a module with same name as package

---

On 28 May 2010 19:36, dsource.org <noreply at dsource.org> wrote:
> phobos commit, revision 1564
>
>
> user: walter
>
> msg:
> bugzilla 4056 Template instantiation with bare parameter not documented
>
> http://www.dsource.org/projects/phobos/changeset/1564
>
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>

Don, what's your top 5 list of unfixed bugs?

>From my personal wish list: I'd really love to see increased focus on
patch integration.  There's a lot of them out there, many with authors who are eager to improve thier patches if needed and just need the nudges. More attention to patches ==> more happy patch submitters ==> more submissions ==> faster bug fixing.

Later,
Brad

On Fri, 28 May 2010, Don Clugston wrote:

> Excellent! That's #1 from my list of the worst bugs.
> 
> There are just two patches I'd like to see in the next release:
> * Brad's patch for arguably the worst wrong-code bug:
> 2127 inliner turns struct "return *this" from by-value into by-ref
> 
> *And this simple compiler fault found by a newbie, which is a very
> recent regression:
> 4242 ICE(module.c): importing a module with same name as package
> 
> ---
> 
> On 28 May 2010 19:36, dsource.org <noreply at dsource.org> wrote:
> > phobos commit, revision 1564
> >
> >
> > user: walter
> >
> > msg:
> > bugzilla 4056 Template instantiation with bare parameter not documented
> >
> > http://www.dsource.org/projects/phobos/changeset/1564
> >
> > _______________________________________________
> > phobos mailing list
> > phobos at puremagic.com
> > http://lists.puremagic.com/mailman/listinfo/phobos
> >
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 

On 28 May 2010 20:34, Brad Roberts <braddr at puremagic.com> wrote:
> Don, what's your top 5 list of unfixed bugs?

After this one:
(A) Deterministic destruction does not work.
3516  Destructor not called on temporaries
3323 Segfault or ICE(e2ir.c) using struct with destructor almost anywhere
1894 scope(exit) is ignored except in compound statements

(B)  Module protection does not work
314 [module] Static, renamed, and selective imports are always public

(C) AAs don't work with structs
2451 Adding structs that use opAssign or postblit to an AA is broken

Which is actually not too different to the list of most voted bugs. These are the ones where I think someone could legitimately decide not to use D based on the existence of that single bug. I used to have a long list of such bugs (most involved wrong code generation), but almost all of them have been fixed.

> From my personal wish list: I'd really love to see increased focus on patch integration. ?There's a lot of them out there, many with authors who are eager to improve their patches if needed and just need the nudges. More attention to patches ==> more happy patch submitters ==> more submissions ==> faster bug fixing.

I agree, and the original success of Linux has often been attributed to that.
We might not be too far from the point where Walter spends most of his
time integrating patches...
And it'll be well worth thinking about what we can do to make that as
easy as possible.

I'll put 3516 on my list to look at this weekend.  1894 is on my list of 'make it die' as well.  Walter, wanna take that one and make it go away for us? :)

Thanks for the list,
Brad

On Fri, 28 May 2010, Don Clugston wrote:

> On 28 May 2010 20:34, Brad Roberts <braddr at puremagic.com> wrote:
> > Don, what's your top 5 list of unfixed bugs?
> 
> After this one:
> (A) Deterministic destruction does not work.
> 3516  Destructor not called on temporaries
> 3323 Segfault or ICE(e2ir.c) using struct with destructor almost anywhere
> 1894 scope(exit) is ignored except in compound statements
> 
> (B)  Module protection does not work
> 314 [module] Static, renamed, and selective imports are always public
> 
> (C) AAs don't work with structs
> 2451 Adding structs that use opAssign or postblit to an AA is broken
> 
> Which is actually not too different to the list of most voted bugs. These are the ones where I think someone could legitimately decide not to use D based on the existence of that single bug. I used to have a long list of such bugs (most involved wrong code generation), but almost all of them have been fixed.
> 
> > From my personal wish list: I'd really love to see increased focus on patch integration. ?There's a lot of them out there, many with authors who are eager to improve their patches if needed and just need the nudges. More attention to patches ==> more happy patch submitters ==> more submissions ==> faster bug fixing.
> 
> I agree, and the original success of Linux has often been attributed to that.
> We might not be too far from the point where Walter spends most of his
> time integrating patches...
> And it'll be well worth thinking about what we can do to make that as
> easy as possible.
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 

Ok.

Brad Roberts wrote:
> I'll put 3516 on my list to look at this weekend.  1894 is on my list of 'make it die' as well.  Walter, wanna take that one and make it go away for us? :)
>
> Thanks for the list,
> Brad
>
> On Fri, 28 May 2010, Don Clugston wrote:
>
> 
>> On 28 May 2010 20:34, Brad Roberts <braddr at puremagic.com> wrote:
>> 
>>> Don, what's your top 5 list of unfixed bugs?
>>> 
>> After this one:
>> (A) Deterministic destruction does not work.
>> 3516  Destructor not called on temporaries
>> 3323 Segfault or ICE(e2ir.c) using struct with destructor almost anywhere
>> 1894 scope(exit) is ignored except in compound statements
>>
>> (B)  Module protection does not work
>> 314 [module] Static, renamed, and selective imports are always public
>>
>> (C) AAs don't work with structs
>> 2451 Adding structs that use opAssign or postblit to an AA is broken
>>
>> Which is actually not too different to the list of most voted bugs. These are the ones where I think someone could legitimately decide not to use D based on the existence of that single bug. I used to have a long list of such bugs (most involved wrong code generation), but almost all of them have been fixed.
>>
>> 
>>> From my personal wish list: I'd really love to see increased focus on
>>> patch integration.  There's a lot of them out there, many with authors who
>>> are eager to improve their patches if needed and just need the nudges.
>>> More attention to patches ==> more happy patch submitters ==> more
>>> submissions ==> faster bug fixing.
>>> 
>> I agree, and the original success of Linux has often been attributed to that.
>> We might not be too far from the point where Walter spends most of his
>> time integrating patches...
>> And it'll be well worth thinking about what we can do to make that as
>> easy as possible.
>> _______________________________________________
>> phobos mailing list
>> phobos at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/phobos
>> 
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> phobos mailing list
>> phobos at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/phobos

Excellent.  And if I might suggest 2 more goals for the next release:

  1) close out at least 10 bugs that have patches
  2) close out at least 5 bugs marked as regressions

to find #1:
has a patch attached to the report (91 issues):
http://d.puremagic.com/issues/buglist.cgi?order=Importance&field0-0-0=attachments.ispatch&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&type0-0-0=equals&value0-0-0=1&product=D

has the patch keyword (113 issues): http://d.puremagic.com/issues/buglist.cgi?keywords=patch&keywords_type=allwords&order=Importance&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D

to find #2 (34 issues): http://d.puremagic.com/issues/buglist.cgi?order=Importance&bug_severity=regression&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D

Hopeful,
Brad

On Fri, 28 May 2010, Walter Bright wrote:

> Date: Fri, 28 May 2010 17:41:26 -0700
> From: Walter Bright <walter at digitalmars.com>
> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
> To: Discuss the phobos library for D <phobos at puremagic.com>
> Subject: Re: [phobos] phobos commit, revision 1564
> 
> Ok.
> 
> Brad Roberts wrote:
> > I'll put 3516 on my list to look at this weekend.  1894 is on my list of 'make it die' as well.  Walter, wanna take that one and make it go away for us? :)
> > 
> > Thanks for the list,
> > Brad
> > 
> > On Fri, 28 May 2010, Don Clugston wrote:
> > 
> > 
> > > On 28 May 2010 20:34, Brad Roberts <braddr at puremagic.com> wrote:
> > > 
> > > > Don, what's your top 5 list of unfixed bugs?
> > > > 
> > > After this one:
> > > (A) Deterministic destruction does not work.
> > > 3516  Destructor not called on temporaries
> > > 3323 Segfault or ICE(e2ir.c) using struct with destructor almost anywhere
> > > 1894 scope(exit) is ignored except in compound statements
> > > 
> > > (B)  Module protection does not work
> > > 314 [module] Static, renamed, and selective imports are always public
> > > 
> > > (C) AAs don't work with structs
> > > 2451 Adding structs that use opAssign or postblit to an AA is broken
> > > 
> > > Which is actually not too different to the list of most voted bugs. These are the ones where I think someone could legitimately decide not to use D based on the existence of that single bug. I used to have a long list of such bugs (most involved wrong code generation), but almost all of them have been fixed.
> > > 
> > > 
> > > > From my personal wish list: I'd really love to see increased focus on
> > > > patch integration.  There's a lot of them out there, many with authors
> > > > who
> > > > are eager to improve their patches if needed and just need the nudges.
> > > > More attention to patches ==> more happy patch submitters ==> more
> > > > submissions ==> faster bug fixing.
> > > > 
> > > I agree, and the original success of Linux has often been attributed to
> > > that.
> > > We might not be too far from the point where Walter spends most of his
> > > time integrating patches...
> > > And it'll be well worth thinking about what we can do to make that as
> > > easy as possible.
> > > _______________________________________________
> > > phobos mailing list
> > > phobos at puremagic.com
> > > http://lists.puremagic.com/mailman/listinfo/phobos
> > > 
> > > ------------------------------------------------------------------------
> > > 
> > > _______________________________________________
> > > phobos mailing list
> > > phobos at puremagic.com
> > > http://lists.puremagic.com/mailman/listinfo/phobos
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 

Ok, and if you want to cheat, there's 5 bugs that are both regressions and have the patch keyword:

http://d.puremagic.com/issues/buglist.cgi?keywords=patch&keywords_type=allwords&order=Importance&bug_severity=regression&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D



On Fri, 28 May 2010, Brad Roberts wrote:

> Date: Fri, 28 May 2010 17:57:09 -0700 (PDT)
> From: Brad Roberts <braddr at puremagic.com>
> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
> To: Discuss the phobos library for D <phobos at puremagic.com>
> Subject: Re: [phobos] phobos commit, revision 1564
> 
> Excellent.  And if I might suggest 2 more goals for the next release:
> 
>   1) close out at least 10 bugs that have patches
>   2) close out at least 5 bugs marked as regressions
> 
> to find #1:
> has a patch attached to the report (91 issues):
> http://d.puremagic.com/issues/buglist.cgi?order=Importance&field0-0-0=attachments.ispatch&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&type0-0-0=equals&value0-0-0=1&product=D
> 
> has the patch keyword (113 issues): http://d.puremagic.com/issues/buglist.cgi?keywords=patch&keywords_type=allwords&order=Importance&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D
> 
> to find #2 (34 issues): http://d.puremagic.com/issues/buglist.cgi?order=Importance&bug_severity=regression&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D
> 
> Hopeful,
> Brad
> 
> On Fri, 28 May 2010, Walter Bright wrote:
> 
> > Date: Fri, 28 May 2010 17:41:26 -0700
> > From: Walter Bright <walter at digitalmars.com>
> > Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
> > To: Discuss the phobos library for D <phobos at puremagic.com>
> > Subject: Re: [phobos] phobos commit, revision 1564
> > 
> > Ok.
> > 
> > Brad Roberts wrote:
> > > I'll put 3516 on my list to look at this weekend.  1894 is on my list of 'make it die' as well.  Walter, wanna take that one and make it go away for us? :)
> > > 
> > > Thanks for the list,
> > > Brad
> > > 
> > > On Fri, 28 May 2010, Don Clugston wrote:
> > > 
> > > 
> > > > On 28 May 2010 20:34, Brad Roberts <braddr at puremagic.com> wrote:
> > > > 
> > > > > Don, what's your top 5 list of unfixed bugs?
> > > > > 
> > > > After this one:
> > > > (A) Deterministic destruction does not work.
> > > > 3516  Destructor not called on temporaries
> > > > 3323 Segfault or ICE(e2ir.c) using struct with destructor almost anywhere
> > > > 1894 scope(exit) is ignored except in compound statements
> > > > 
> > > > (B)  Module protection does not work
> > > > 314 [module] Static, renamed, and selective imports are always public
> > > > 
> > > > (C) AAs don't work with structs
> > > > 2451 Adding structs that use opAssign or postblit to an AA is broken
> > > > 
> > > > Which is actually not too different to the list of most voted bugs. These are the ones where I think someone could legitimately decide not to use D based on the existence of that single bug. I used to have a long list of such bugs (most involved wrong code generation), but almost all of them have been fixed.
> > > > 
> > > > 
> > > > > From my personal wish list: I'd really love to see increased focus on
> > > > > patch integration.  There's a lot of them out there, many with authors
> > > > > who
> > > > > are eager to improve their patches if needed and just need the nudges.
> > > > > More attention to patches ==> more happy patch submitters ==> more
> > > > > submissions ==> faster bug fixing.
> > > > > 
> > > > I agree, and the original success of Linux has often been attributed to
> > > > that.
> > > > We might not be too far from the point where Walter spends most of his
> > > > time integrating patches...
> > > > And it'll be well worth thinking about what we can do to make that as
> > > > easy as possible.
> > > > _______________________________________________
> > > > phobos mailing list
> > > > phobos at puremagic.com
> > > > http://lists.puremagic.com/mailman/listinfo/phobos
> > > > 
> > > > ------------------------------------------------------------------------
> > > > 
> > > > _______________________________________________
> > > > phobos mailing list
> > > > phobos at puremagic.com
> > > > http://lists.puremagic.com/mailman/listinfo/phobos
> > _______________________________________________
> > phobos mailing list
> > phobos at puremagic.com
> > http://lists.puremagic.com/mailman/listinfo/phobos
> > 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
> 

Moving this over to dmd-internals since it's not phobos related any more..

Another old bug I remembered on the drive home tonight:
  http://d.puremagic.com/issues/show_bug.cgi?id=1513
  try/catch/finally misbehavior on windows

I just updated it with the current test results.. still different behavior with exception handling between windows and linux which is pretty bad.

I've never looked at the exception handing on windows (and would prefer not to start :).  The linux behavior is the one I'd call correct.

- Brad

On 5/28/2010 6:01 PM, Brad Roberts wrote:
> Ok, and if you want to cheat, there's 5 bugs that are both regressions and have the patch keyword:
> 
> http://d.puremagic.com/issues/buglist.cgi?keywords=patch&keywords_type=allwords&order=Importance&bug_severity=regression&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D
> 
> 
> 
> On Fri, 28 May 2010, Brad Roberts wrote:
> 
>> Date: Fri, 28 May 2010 17:57:09 -0700 (PDT)
>> From: Brad Roberts <braddr at puremagic.com>
>> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
>> To: Discuss the phobos library for D <phobos at puremagic.com>
>> Subject: Re: [phobos] phobos commit, revision 1564
>>
>> Excellent.  And if I might suggest 2 more goals for the next release:
>>
>>   1) close out at least 10 bugs that have patches
>>   2) close out at least 5 bugs marked as regressions
>>
>> to find #1:
>> has a patch attached to the report (91 issues):
>> http://d.puremagic.com/issues/buglist.cgi?order=Importance&field0-0-0=attachments.ispatch&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&type0-0-0=equals&value0-0-0=1&product=D
>>
>> has the patch keyword (113 issues): http://d.puremagic.com/issues/buglist.cgi?keywords=patch&keywords_type=allwords&order=Importance&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D
>>
>> to find #2 (34 issues): http://d.puremagic.com/issues/buglist.cgi?order=Importance&bug_severity=regression&query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=D
>>
>> Hopeful,
>> Brad
>>
>> On Fri, 28 May 2010, Walter Bright wrote:
>>
>>> Date: Fri, 28 May 2010 17:41:26 -0700
>>> From: Walter Bright <walter at digitalmars.com>
>>> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
>>> To: Discuss the phobos library for D <phobos at puremagic.com>
>>> Subject: Re: [phobos] phobos commit, revision 1564
>>>
>>> Ok.
>>>
>>> Brad Roberts wrote:
>>>> I'll put 3516 on my list to look at this weekend.  1894 is on my list of 'make it die' as well.  Walter, wanna take that one and make it go away for us? :)
>>>>
>>>> Thanks for the list,
>>>> Brad
>>>>
>>>> On Fri, 28 May 2010, Don Clugston wrote:
>>>>
>>>> 
>>>>> On 28 May 2010 20:34, Brad Roberts <braddr at puremagic.com> wrote:
>>>>> 
>>>>>> Don, what's your top 5 list of unfixed bugs?
>>>>>> 
>>>>> After this one:
>>>>> (A) Deterministic destruction does not work.
>>>>> 3516  Destructor not called on temporaries
>>>>> 3323 Segfault or ICE(e2ir.c) using struct with destructor almost anywhere
>>>>> 1894 scope(exit) is ignored except in compound statements
>>>>>
>>>>> (B)  Module protection does not work
>>>>> 314 [module] Static, renamed, and selective imports are always public
>>>>>
>>>>> (C) AAs don't work with structs
>>>>> 2451 Adding structs that use opAssign or postblit to an AA is broken
>>>>>
>>>>> Which is actually not too different to the list of most voted bugs. These are the ones where I think someone could legitimately decide not to use D based on the existence of that single bug. I used to have a long list of such bugs (most involved wrong code generation), but almost all of them have been fixed.
>>>>>
>>>>> 
>>>>>> From my personal wish list: I'd really love to see increased focus on
>>>>>> patch integration.  There's a lot of them out there, many with authors
>>>>>> who
>>>>>> are eager to improve their patches if needed and just need the nudges.
>>>>>> More attention to patches ==> more happy patch submitters ==> more
>>>>>> submissions ==> faster bug fixing.
>>>>>> 
>>>>> I agree, and the original success of Linux has often been attributed to
>>>>> that.
>>>>> We might not be too far from the point where Walter spends most of his
>>>>> time integrating patches...
>>>>> And it'll be well worth thinking about what we can do to make that as
>>>>> easy as possible.
>>>>> _______________________________________________

On 29 May 2010 02:57, Brad Roberts <braddr at puremagic.com> wrote:
> Excellent. ?And if I might suggest 2 more goals for the next release:
>
> ?1) close out at least 10 bugs that have patches

Since the goal seems to be to encourage community involvement, I'd
change that to:
close bugs with patches by at least 5 different authors. Most the last
dozen releases have had ten of my patches included, I don't think
anyone will be terribly excited if that happens again. I think there's
already 3 authors represented in the next release.
I suggest these bugs which look pretty straightforward (and therefore
unlikely to cause regressions):
3651 mangleof broken for enums   (by mpiepk)
4084 Ignored missing main() closing bracket  (ellery newcomer)
2881 x.stringof returns typeof(x).stringof when x is an enum (nfxjfg)
3854 Error on static initialization of arrays with trailing comma.
(regression, ibuclaw/ellery newcomer)
3554 for option -od for relative path the path is added twice
(regression, thecybershadow)

> ?2) close out at least 5 bugs marked as regressions

1193 regression: "matches more than one template declaration" doesn't list the location of the conflicting templates  (bradr)

is the only other regression with an up-to-date, trouble-free patch. So I think that's a tough ask.

But really, the patch for the inliner bug 2127 is the most important.