4 ms·

I think the author is mischaracterizing all of these examples; none of them are "second systems" but major revisions of software with large install bases and no

by choult 11y ago

I think the author is mischaracterizing all of these examples; none of them are "second systems" but major revisions of software with large install bases and no built-in update mechanism. They are reliant on distributions and third parties to provide any form of automatic update; oftentimes the upgrade path without distribution-level support is painful for an average user. Certainly, in the case of PHP, it was a very painful process to get major adoption of 5 - especially as so many of the users are reliant on managed hosting. It took a long time for the install base to grow there; PHP 7 will similarly take a long time to get up there. At least the decision to scrap 6 (a decision also taken in light of all of the literature out there describing the features of 6) was a pragmatic one; PHP 7 is now just around the corner. What the author is describing are two general problems for any software development where there is a large install-base: development cycles and release management. Just look at uptake numbers for operating systems for an example as to why this is absolutely not a problem confined to scripting languages or even OSS software: http://en.wikipedia.org/wiki/Usage_share_of_operating_systems http://en.wikipedia.org/wiki/Usage_share_of_operating_system...

cygx 11y ago
The author does mention Second Deployment Syndrome. And Perl6 most definitely is a 'second system'.
castell 11y ago
The transition from PHP3 to PHP4 and from PHP4 to PHP5 wasn't that painful. At the time, most PHP open source projects simply avoided OO concepts due to their slow implementation in PHP4 and to be downwards compatible with PHP3. It solved the transition as they only slowly adopted class based OO later in PHP5 era. Also the transition period of parallel version of PHP4 and 5 was between 2004 and 2008 - long enough so that most shared hosters were able to make the switch. PHP6 failed, and most planned features were backported to PHP5.3+. As several books were published in the months before the planned PHP6 release, the next PHP version will be named PHP7 to avoid any confusion. Nevertheless, PHP7 isn't a bigger change than let's say PHP 5.2 to 5.3. Overall I would say, PHP's developers and the community has done a great job. Modern PHP.net and HHVM are certainly better than their reputation. The same goes for MySQL and MariaDB.
- Macha 11y ago
  PHP4->PHP5 required the gophp5 campaign in the end (domain now occupied by squatters) where several open source projects (iirc, Drupal, WordPress and phpBB among them) committed to dropping php4 support to pressure shared hosts into upgrading. And that wasn't even a backwards incompatible change (unless you were using things already deprecated and disabled by default at the end of php4, like register_globals) Imagine if in 2-3 years, SQLAlchemy, Requests and SciPy for example dropped Python 2 which would be a similar Python attempt (not saying they would or should). That's some pretty forceful migration.
  - stephenr 11y ago
    One small (afaik) correction: Wordpress was never part of the GoPHP5 initiative. Wordpress has a terrible history of maintaining backwards compatibility over good design, and modernising their codebase. They only started to add support for a modern DB driver (mysqli) a bit over a year ago. MySQLi and PDO have been in PHP since v5.0 and v5.1, respectively, that's 11 and 10 years ago.
- Bahamut 11y ago
  If anything, my takeaway here is that PHP might have handled things the best of the three. PHP doesn't have the huge schism that Python 3 does, and the development cycle isn't nearly as crazy as Perl. Sometimes pragmatism is more important than idealism.
  - wvenable 11y ago
    The process from PHP4 to PHP5 was long but there were clear advantages to PHP5 that it made it logical choice for all new projects. I'm still working with brand new Python 2.7 code.
  - juliangregorian 11y ago
    We don't know that, because PHP 7 isn't out yet, but also PHP 7 doesn't change in breaking ways. All of the language warts that plagued 4 and 5 are still around.
  - 11y ago
pornel 11y ago
I'd say PHP's major upgrades were a success. Adoption of PHP5 was a pain, but still it was quicker than adoption of Python 3. At least the community was all behind PHP5 and there was a major push towards it. I think PHP has dodged the bullet by taking PHP6 apart and reintroducing features piecemeal in PHP 5.3+.

by choult 11y ago

cygx 11y ago

The author does mention Second Deployment Syndrome. And Perl6 most definitely is a 'second system'.

castell 11y ago

The transition from PHP3 to PHP4 and from PHP4 to PHP5 wasn't that painful. At the time, most PHP open source projects simply avoided OO concepts due to their slow implementation in PHP4 and to be downwards compatible with PHP3. It solved the transition as they only slowly adopted class based OO later in PHP5 era. Also the transition period of parallel version of PHP4 and 5 was between 2004 and 2008 - long enough so that most shared hosters were able to make the switch. PHP6 failed, and most planned features were backported to PHP5.3+. As several books were published in the months before the planned PHP6 release, the next PHP version will be named PHP7 to avoid any confusion. Nevertheless, PHP7 isn't a bigger change than let's say PHP 5.2 to 5.3. Overall I would say, PHP's developers and the community has done a great job. Modern PHP.net and HHVM are certainly better than their reputation. The same goes for MySQL and MariaDB.

Macha 11y ago
PHP4->PHP5 required the gophp5 campaign in the end (domain now occupied by squatters) where several open source projects (iirc, Drupal, WordPress and phpBB among them) committed to dropping php4 support to pressure shared hosts into upgrading. And that wasn't even a backwards incompatible change (unless you were using things already deprecated and disabled by default at the end of php4, like register_globals) Imagine if in 2-3 years, SQLAlchemy, Requests and SciPy for example dropped Python 2 which would be a similar Python attempt (not saying they would or should). That's some pretty forceful migration.
- stephenr 11y ago
  One small (afaik) correction: Wordpress was never part of the GoPHP5 initiative. Wordpress has a terrible history of maintaining backwards compatibility over good design, and modernising their codebase. They only started to add support for a modern DB driver (mysqli) a bit over a year ago. MySQLi and PDO have been in PHP since v5.0 and v5.1, respectively, that's 11 and 10 years ago.
Bahamut 11y ago
If anything, my takeaway here is that PHP might have handled things the best of the three. PHP doesn't have the huge schism that Python 3 does, and the development cycle isn't nearly as crazy as Perl. Sometimes pragmatism is more important than idealism.
- wvenable 11y ago
  The process from PHP4 to PHP5 was long but there were clear advantages to PHP5 that it made it logical choice for all new projects. I'm still working with brand new Python 2.7 code.
- juliangregorian 11y ago
  We don't know that, because PHP 7 isn't out yet, but also PHP 7 doesn't change in breaking ways. All of the language warts that plagued 4 and 5 are still around.
- 11y ago

pornel 11y ago

I'd say PHP's major upgrades were a success. Adoption of PHP5 was a pain, but still it was quicker than adoption of Python 3. At least the community was all behind PHP5 and there was a major push towards it. I think PHP has dodged the bullet by taking PHP6 apart and reintroducing features piecemeal in PHP 5.3+.