On 7/2/07, Scott MacVicar <[EMAIL PROTECTED]> wrote:
Removing non stable PECL extensions would be a good step. I had a quick
glance through what is currently there and as far as I can see there is
only a handful of extensions I'd include:
* APC
* big_int
* htscanner
* Imagick
* memcache
* pecl_http
* phar
* SPL_Types
* tidy
If an author would like his extension in the Windows installer then they
could just ask, would prevent unmaintained and unstable extensions
included in the build.
But we are shipping them currently in the zip build ( all I'm doing is
repackaging php-5.2.x-win32.zip and pecl-5.2.x-win32.zip ), so I think
the issue would be better dealt with at the PECL level. We should have
some sort of stability status similar to PEAR and use that when
determining which extensions to ship.
Could the features potentially be grouped into two trees? Core and added
functionality that can come from PECL?
Originally, I was thinking the same thing, but the consensus was to
make it the way it is currently. I like the two tree approach myself
and switch to that if that what everyone wants to do. But I do
remember the disention is that it seemed to confusing; perhaps
splitting them up based upon stability instead?
Scott
John Mertic wrote:
> On 2 Jul 2007 09:30:13 -0000, [EMAIL PROTECTED]
>> The Windows Installer seems to allow someone to enable just about every
>> extension that can be built, roughly about 118. It seems that most
>> people assume that they need every feature of PHP to work, especially
>> thsoe in Windows.
>>
>> Out of those 118 extensions there are 3 which cause crashes and blank
>> pages.
>>
>> SAM extension has some interesting code its its RINIT function to
>> require a PHP file, if its not there it exits and you get a blank page
>> with display errors off.
>>
>> The thread pecl extension and iisfunc don't work when and cause crashes
>> on module shutdown.
>>
>> I propose that we disable all experimental extensions in the installer
>> to prevent this, some of these extensions don't even have a release
>> listed on http://pecl.php.net
>>
>> Scott
>
> My original intention in creating the installer was to provide a tool
> for easily installing and updating PHP on the Windows platform, with
> as little as possible having to deal with configuration in the php.ini
> file, the web server, or the Windows environment in general. I think
> this is generally true ( or at least haven't heard otherwise ;-> ).
>
> One of the other goals, which was a shortcoming of the previous
> installer, was to provide a way for installing any of the available
> extensions whether from the core or PECL as easy as possible, which is
> true today ( select as feature in the installer, and it's enabled in
> the php.ini file as well ). I'm not sure that getting rid of that
> ability is an improvement; it does get exposure to several PECL
> packages that may otherwise be ignored ( such as APC, which I use a
> lot ).
>
> It sounds to me that there are a few issues going on here:
>
> - The presentation of extensions that are experimental are listed
> alongside those which are relatively stable. That could be fixed by
> putting those in a separate menu, but we would also need to have some
> sort of criteria for each ( Perhaps everything from PECL would be
> seperated ).
> - Several of the extensions require third party dlls that are not
> included with the installer. If we know which ones these are we could
> either include them in the distribution or provide some sort of check
> for their registered presence before allowing them to be installed.
> - The ease of allowing all extensions to be installed. This is mostly
> a fault of the feature tree in MSI itself ( I can't disable that from
> the menu ). I can do a bit of searching to see if any solutions have
> been found to this, but I suspect not.
>
> If we could address the first two items and see what can be done to
> improve the last one, do you think this helps things out?
>
--
--
John Mertic "Explaining a joke
is like dissecting a frog: you
[EMAIL PROTECTED] understand it better,
but the frog dies in the
process."
-Mark Twain
--
PHP Internals - PHP Runtime Development Mailing List
To unsubscribe, visit: http://www.php.net/unsub.php
