May 31, 2013
Dicebot, el 31 de May a las 13:44 me escribiste:
> On Friday, 31 May 2013 at 09:08:17 UTC, Leandro Lucarella wrote:
> >This is just plain and completely wrong. I don't know many big-ish
> >opensource projects that doesn't have release candidates, and I
> >haven't
> >see any "distribution" targeted at end users using release
> >candidates.
> >
> >Have you ever see a Linux distribution shipping an rc kernel (that
> >is
> >not only installable by explicit user action) for example?
> 
> Oh, I have meant it completely other way around - if some release is made available through common channels it does not matter if it is called "beta" or "RC", people will just start using it. Remember the issue with UDA syntax?

The UDAs issue was completely different, there were no betas including UDAs. People using it were just using a development snapshot.

> In mature projects RC does not differ that much from actual release other than by extra regression fixes. But for D process is not THAT smooth enough and it will take some time to settle things down.

This is pretty much how it is now. Only minor regressions can be found in a beta/rc usually. There are no changes in behaviour or new features.

-- 
Leandro Lucarella (AKA luca)                     http://llucax.com.ar/
----------------------------------------------------------------------
GPG Key: 5F5A8D05 (F8CD F9A7 BF00 5431 4145  104C 949E BFB6 5F5A 8D05)
----------------------------------------------------------------------
All fathers are intimidating. They're intimidating because they are
fathers.  Once a man has children, for the rest of his life, his
attitude is, "To hell with the world, I can make my own people. I'll eat
whatever I want. I'll wear whatever I want, and I'll create whoever
I want."
	-- Jerry Seinfeld
May 31, 2013
On Friday, 31 May 2013 at 14:08:18 UTC, Leandro Lucarella wrote:
>> In mature projects RC does not differ that much from actual release
>> other than by extra regression fixes. But for D process is not THAT
>> smooth enough and it will take some time to settle things down.
>
> This is pretty much how it is now. Only minor regressions can be found
> in a beta/rc usually. There are no changes in behaviour or new features.

Erm, I remember you taking good part in const initialization discussion with all semantics changes and compiler flags added until final decision was set in stone. That is something better done in semi-closed beta in my opinion.
May 31, 2013
On Friday, 31 May 2013 at 14:08:17 UTC, Leandro Lucarella wrote:
> And I don't mean to minimize the incredible breakthrough concerning the
> release process in this cycle, just pointing out places were we can
> still do better :)

Btw, I have included minor version number into Arch Linux package version, may suggest other packagers to do the same. Version string shown by DMD front-end itself is not that important as spec shouldn't change within minor versions. Still may be useful though.
May 31, 2013
On Friday, May 31, 2013 10:17:07 Leandro Lucarella wrote:
> Yeah, and that's exactly what I suggested here several times, and ultimately at DConf :). A step forward has been made in this release, as you said, betas were announced in this NG for the first time, before they were announced only in the beta ML.

http://d.puremagic.com/issues/show_bug.cgi?id=10153 http://d.puremagic.com/issues/show_bug.cgi?id=10154

- Jonathan M Davis
May 31, 2013
Dicebot, el 31 de May a las 16:21 me escribiste:
> On Friday, 31 May 2013 at 14:08:17 UTC, Leandro Lucarella wrote:
> >And I don't mean to minimize the incredible breakthrough
> >concerning the
> >release process in this cycle, just pointing out places were we
> >can
> >still do better :)
> 
> Btw, I have included minor version number into Arch Linux package version, may suggest other packagers to do the same. Version string shown by DMD front-end itself is not that important as spec shouldn't change within minor versions. Still may be useful though.

For users it is. I want to know if the compiler I'm used is the latest with all critical bugfixes included or not. Remember that if we are going massive, we can't count anymore on user installing their compilers themselves anymore. People will start just doing an apt-get install dmd, or even have it preinstalled, and it should be easy for them to know exactly what release they are using.

-- 
Leandro Lucarella (AKA luca)                     http://llucax.com.ar/
----------------------------------------------------------------------
GPG Key: 5F5A8D05 (F8CD F9A7 BF00 5431 4145  104C 949E BFB6 5F5A 8D05)
----------------------------------------------------------------------
Y serán tiempos de vanos encuentros entre humano y humano; en que las
fieras se comerán entre ellas y después del final; en que se abríran las
tierras y los cielos... y en el medio de la nada Racing saldrá campeón.
	-- Ricardo Vaporeso
May 31, 2013
Dicebot, el 31 de May a las 16:18 me escribiste:
> On Friday, 31 May 2013 at 14:08:18 UTC, Leandro Lucarella wrote:
> >>In mature projects RC does not differ that much from actual
> >>release
> >>other than by extra regression fixes. But for D process is not
> >>THAT
> >>smooth enough and it will take some time to settle things down.
> >
> >This is pretty much how it is now. Only minor regressions can be found in a beta/rc usually. There are no changes in behaviour or new features.
> 
> Erm, I remember you taking good part in const initialization discussion with all semantics changes and compiler flags added until final decision was set in stone. That is something better done in semi-closed beta in my opinion.

Well, that case could really be considered just a regression, something that used to work one way was changed in the beta (but was wrong) and during the beta process was restored (with a better migration/deprecation plan). I still think is quite different from introducing new features or behaviour changes *intentionally*.

You are never covered 100%. But anyway, I'm not against having a first
iteration with less exposure (i.e. targeted only to DMD devels), but
I don't think we even need a "release" for that. Is enough to say in the
MLs "hey, we are starting with the release process, everyone check the
current master and report any problems" and freeze new features merge
from there. One everything is slightly tested and at least the devels
agree the master is in good shape, we can start shipping proper public
release candidates, with the proper changelog, version number, etc.

-- 
Leandro Lucarella (AKA luca)                     http://llucax.com.ar/
----------------------------------------------------------------------
GPG Key: 5F5A8D05 (F8CD F9A7 BF00 5431 4145  104C 949E BFB6 5F5A 8D05)
----------------------------------------------------------------------
- Tata Dios lo creó a usté solamente pa despertar al pueblo y fecundar
  las gayinas.
- Otro constrasentido divino... Quieren que yo salga de joda con las
  hembras y después quieren que madrugue.
	-- Inodoro Pereyra y un gallo
May 31, 2013
On Friday, 31 May 2013 at 14:08:17 UTC, Leandro Lucarella wrote:
>
> About this, AFAIK 2.063.1 is really what's in the release, but the
> binary version number (and the zip name) have only 2.063. I think that
> should be fixed and the real version number should be present in both
> downloadables and binary. Also a micro changelog should be provided,
> only with the regressions that were fixed.

Of course, that absolutely makes sense and should be implemented by next release if possible.

> And I don't mean to minimize the incredible breakthrough concerning the
> release process in this cycle, just pointing out places were we can
> still do better :)

Agreed. Looking back just a couple of releases ago, the situation has improved considerably, but as always there's a lot more improvements that can and should be done.

As for the comment that RC's will be treated as stable releases, that's hard to swallow, esp when you consider what's going on now. The current release is worse than a RC because it's not labeled for what it is, people will think it's stable when in fact it's not. I think that it is far more professional and responsible to explicitly state that the version on the download page is a release candidate rather than not saying anything at all. People will get the wrong impression and think that it is a well tested and honed stable release.

To reduce potential confusion, we can place RC's in a separate download page.

Finally the RC can be a reasonably well tested version that is near completion to minimize the amount of re-work and bug potential. Even if it is misused by people who should know better, it'll still perform reasonably well, and the rest of us tinkerers will greatly benefit from having it.

Finally, making RC's available to the public will greatly help increase the quality of the final product and increase the confidence in it for production use.

It'll be a win-win for everyone, no question in my mind.

--rt
May 31, 2013
On 05/30/2013 08:16 AM, Andrei Alexandrescu wrote:
> Hello,
>
>
> We are pleased to announce that dmd 2.063, the reference compiler of the
> D programming language, is now available for download for OSX, Windows,
> and a variety of Unixen:
>

The rpm package doesn't make the appropriate links in /usr/lib, so when I try to build a shared library, at runtime it issues

./test1d: error while loading shared libraries: libphobos2.so.0.63: cannot open shared object file: No such file or directory

I would assume the deb package has the same shortcoming

May 31, 2013
On Fri, 2013-05-31 at 12:19 -0700, Ellery Newcomer wrote: […]
> I would assume the deb package has the same shortcoming

I have not seen this with the deb on Debian Unstable.

-- 
Russel. ============================================================================= Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net 41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder


May 31, 2013
On 05/31/2013 12:32 PM, Russel Winder wrote:
> On Fri, 2013-05-31 at 12:19 -0700, Ellery Newcomer wrote:
> […]
>> I would assume the deb package has the same shortcoming
>
> I have not seen this with the deb on Debian Unstable.
>

just tried it on ubuntu 12.10, and it does the same.

are you using -defaultlib=libphobos2.so

?