[RFC] Deprecate PEAR/PECL & Replace with composer/pickle

I don't think anyone is arguing that literally no one is using PEAR
packages, but looking at the ~6 months old archive.org mirror of
http://pear.php.net/package-stats.php (available at
https://web.archive.org/web/20160315083551/http://pear.php.net/package-stats.php),
the largest package, PEAR itself only has around 800,000 downloads, and the
fifth largest package, XML_RPC only has around 10,000 downloads.

This can be compared with Packagist, where the most popular package
(psr/log) had approximatly 12,000,000 downloads during the last 6 months,
and the fifth most popular package (symfony/yaml) had approximately
10,000,000 downloads.

PEAR isn't dead, but I don't think it's an exaggeration to say that it's
moribund.

On Tue, Sep 13, 2016 at 10:40 AM, Tony Marston tonymarston@hotmail.com
wrote:

Sent: Monday, September 12, 2016 12:34 PM

To: Tony Marston
Cc: Sherif Ramadan ; Stanislav Malyshev ; internals@lists.php.net
Subject: Re: [PHP-DEV] [RFC] Deprecate PEAR/PECL & Replace with
composer/pickle

On Mon, Sep 12, 2016 at 9:32 AM, Tony Marston tonymarston@hotmail.com
wrote:

Sent: Sunday, September 11, 2016 11:44 AM

To: Tony Marston
Cc: Stanislav Malyshev ; internals@lists.php.net
Subject: Re: [PHP-DEV] [RFC] Deprecate PEAR/PECL & Replace with
composer/pickle

On Sun, Sep 11, 2016 at 4:47 AM, Tony Marston tonymarston@hotmail.com
wrote:

Sent: Saturday, September 10, 2016 7:47 PM

To: Tony Marston ; internals@lists.php.net
Subject: Re: [PHP-DEV] [RFC] Deprecate PEAR/PECL & Replace with
composer/pickle

Then I suggest that those who are so anxious to see of death of
PEAR/PECL should be forced to provide a viable alternative first. Otherwise
they would be just like those stupid politicians who try to force commuters
out of their private cars and into public transport without realising that
the existing public transport system is NOT a viable replacement and is
incapable of taking on the extra load.

I just want to say that PEAR as a source repository, has been dead for
quite some time. It's filled with outdated code that has hardly seen any
maintenance in years, and nobody really contributes to it anyway.

PEAR/PECL as a package manager has historically had little utility to
the average user apart from installing those PECL extensions which aren't
packaged by a particular user's distribution repository. Certainly hasn't
had any real viability in years. Trying to replace something that's
inherently non-viable with a viable-alternative seems like a pretty moot
point.

Just because you have found no use for it does not mean that others feel
the same. How about those large numbers of websites that have to use PEAR
Mailer instead of the built-in mail() function? I personally use SVNManager
to manage my SVN repositories, and this is dependent on one of the PEAR
modules, so there is a very recent use. How many other PEAR modules have
been installed and are still is use? Do you have the download statistics
for all the PEAR modules? That would be much more accurate that all this
guesswork and supposition.

--
Tony Marston

we have download statistics for pear/pecl:
http://pecl.php.net/package-stats.php
http://pear.php.net/package-stats.php
but it would be better to filter the stats to a more recent time interval
like last year or something, not sure if that is possible with the current
web interface, but you can see the recent download stats on the package
page, eg:
http://pecl.php.net/package-stats.php?cid=33&pid=876

http://pear.php.net/package-stats.php?pid=14&cid=19

Those download stats STILL show multiple downloads for multiple packages
in 2016, so that puts paid to the idea that nobody uses it. Those who keep
saying “none of the people I know use it” inhabit tiny groups who are NOT
indicative of the entire PHP ecosystem.

--
Tony Marston