RE: [INTERNALS-WIN] RE: Compilers, SDK support roadmap

That sounds reasonable although we don't have a track record building
with VC9. But we are probably better off trying to adopt what will be
more aligned with what's coming down the pipe in the next two years than
staying on an older version.
Regardless I wouldn't want to confuse people with supporting both VC8
and VC9. We should have only two with the hope of completely dropping
VC6 as soon as possible.

Andi


> -----Original Message-----
> From: Garrett Serack [mailto:[email protected]]
> Sent: Wednesday, July 16, 2008 8:06 AM
> To: Pierre Joye; Elizabeth M Smith
> Cc: [email protected]
> Subject: RE: [INTERNALS-WIN] RE: Compilers, SDK support roadmap
> 
> Sorry for joining the party late...
> 
> I think for the most part, I'm happy with the VC6/VC8 plan.
> 
> Before completely committing to that path, I think that we should
> review
> a couple of quick thoughts I had about this over the weekend.
> 
>    1. The differences between VC8 and VC9 are pretty small--compared
to
>    well, any two other versions of VC.
> 
>    2. VC9 does provide the 'correct' environment, libraries, headers
> and
>    whatnot for Server 2008. Code from other versions of VC obviously
> run
>    on Server 2008, but VC9 includes support for new libraries and
> technology
>    specifically for newer OSes.
> 
>    3. "Windows 7" will be a mid-band release, and significant kernel
> changes
>    are not part of mid-band releases. VC9 will be the favored platform
> for
>    this whole cycle, and the cycle next. And given Stephen Sinofski's
>    desire to ship Windows releases ever two years, we can rely on that
> VC9
>    will be the solidified platform for an extended period of time.
> 
>    4. The Windows SDK for Server 2008 (v6.1) does come with all the
> tools,
>    and all are licensed for commercial use freely--including MASM,
> which for
>    MASM 8.0 (if you want to use VC8), is a separate download, and even
> then
>    the EULA is non-commercial -- a situation I'm trying to rectify.
> With
>    the 6.1 Windows SDK, it's a one-stop-shopping affair for compilers,
> code,
>    libraries, and tools. (and it's one file!--rather than the 15 files
> for
>    to download the previous version's offline install).
> 
>    Heck, even then you need Visual Studio 2005 Express, two SP's (one
> for VS
>    and one for VS for Vista). the VC8 platform is a Tooling nightmare.
> Oh,
>    and don't install them  in the wrong order (like SDK then VS... no-
> worky)
> 
>    5. The stability of VC9 is pretty damn good. I know that a lot of
> product
>    groups are relying on it, and the entire build of "Windows 7" is
all
> using
>    that right now.
> 
>    6. I can work with apache to go straight to VC9 for the same
> reasons. Really
>    if it works on VC8, VC9 is not a problem.  Given that this is new
> territory
>    anyway, maybe getting as close to the present is a good idea.  I'm
> working
>    with the folks from Apache, and I can easily push for VC9 on the
> same basis.
>    Frankly, this is probably far easier anyway.
> 
>    7. The tool model/layout for the 6.1 SDK is expected to be far more
>    consistent going forward; so upgrading to the next SDK generation
> should be
>    far easier.
> 
> Frankly, I'm now starting to lean to the VC6/VC9 model, with VC6
> dropping post
> 5.3.  I really can't say that I see any VC9 adoption blockers, and
it's
> significantly easier to setup the build environment.
> 
> Garrett
> 
> 
> -----Original Message-----
> From: Pierre Joye [mailto:[email protected]]
> Sent: Wednesday, July 16, 2008 7:35 AM
> To: Elizabeth M Smith
> Cc: [email protected]
> Subject: Re: [INTERNALS-WIN] RE: Compilers, SDK support roadmap
> 
> hi!
> 
> On Wed, Jul 16, 2008 at 4:18 PM, Elizabeth M Smith
> <[email protected]> wrote:
> > If support is the issue, we can wait on 9, doesn't matter too much
to
> me
> > as long as VC6 starts the death march.
> >
> > However I think we need to NOT do any VC9 builds until VC6 builds
are
> > discontinued - supporting two compiler versions for a release will
> > strain resources already.
> 
> Given than 5.3 is the last release to use VC6, it will certainly not
> happen. The last line about VC8/9 (x86/x64) is what it seems to be
> confusing. The idea here was to provide only VC8 and VC9 builds, not
> VC6. But I'm 99.99% sure that we will have a 5.4, that will simply
> kill VC6 sooner =).
> 
> 
> > I'd say VC6 and VC8 "official" builds for 5.3, plus VC8-64 bit
> > experimental builds (until they work well).
> 
> That's what my proposal says, good :)
> 
> > Then for PHP 5.4 or 6.0 (whichever comes first) kill off VC6 support
> and
> > add VC9 and VC9-64bit.
> 
> Ditto
> 
> --
> Pierre
> http://blog.thepimp.net | http://www.libgd.org
> 
> --
> Windows Internals Mailing List (http://www.php.net/)
> To unsubscribe, visit: http://www.php.net/unsub.php
> 
> 
> 
> --
> Windows Internals Mailing List (http://www.php.net/)
> To unsubscribe, visit: http://www.php.net/unsub.php