Re: [INTERNALS-WIN] end of support for VC6 in trunk
| From: | Pierre Joye | Date: | Tue, 22 Sep 2009 20:43:18 +0000 |
| Subject: | Re: [INTERNALS-WIN] end of support for VC6 in trunk | ||
| References: | 1 2 3 4 | Groups: | php.internals.win |
| Request: | Send a blank email to internals-win+get-620@lists.php.net to get a copy of this message | ||
Hi!
On Tue, Sep 22, 2009 at 10:21 PM, William A. Rowe, Jr.
<wrowe@rowe-clan.net> wrote:
> Pierre Joye wrote:
>> On Tue, Sep 22, 2009 at 9:53 PM, Guilherme Blanco
>> <guilhermeblanco@gmail.com> wrote:
>>> Are there any feedback from Apache Group regarding this end of support?
>>>
>>> Official Apache packages are still compiled with VC6, and it's needed
>>> to download from ApacheLounge to pick builds of VC9.
>>
>> The next major version will stop to support vc6, as far as I can tell.
>> But even if there is no official statement
>
> I don't think I've made an official statement, because ASF binary support
> isn't official, it's incidental on a committer-volunteer basis (and no,
> outside binary artifacts are not accepted as network-of-trust policy).
> Although someone could still ship VC6 compiled artifacts, I'll be offering
> only VC10 (or whatever the flavor-of-the-week is) for that next version.
>
> But it's altogether irrelevant, and PHP key leaders have all stated that
> FastCGI is the 'supported'[1] method of interop;
I don't think we ever did that, except for IIS.
> as of the next beta of
> ASF mod_fcgid, there will be a binary as well for win32, and as of 2.4/3.0
> httpd release, mod_fcgid or mod_proxy_fcgi will likely be part of the
> default distribution. While it's nice to run both as 32 or 64 bit, even
> that isn't strictly required. Compilation interop issues should be gone.
That's a very good news :)
> Since the whole mod_php SetIni is not threadsafe,
What are you refering to here? :)
> and can't be (the Zend
> API is insufficient w/zts enabled), and only nominal efforts were made at
> making PHP a thread-supporting, multiple-application host VM, it has
> relegated to the single-task::single-process architecture of the 80's.
I'd to agree here, even if the engine itself is not really the problem
but the dependencies.
Cheers,
--
Pierre
http://blog.thepimp.net | http://www.libgd.org