The POSIX regex functions are currently marked as DEPRECATED in 5.3.0 and are due to be removed completely in PHP 6. I propose that instead of being dropped they be converted into wrappers for the equivalent PCRE functions. It does not matter how long various people have been preaching in various unofficial channels that users should switch to using the PCRE functions (preg_*) instead of the POSIX ones (ereg_*) as many users simply do not access those channels. The first time that a significant number of users will become aware of this is when they switch to 5.3.0 or later, and they will be very, very upset. Some users do not test their applications against each new PHP version as it comes out, they just use whatever version their hosting company provides them with. It does not matter that the PHP developers cannot find someone to upgrade the POSIX functions, it does not matter that the same functionality can be provided with the PCRE functions, all the users will know is that their scripts, some of which have been running for many years, will suddenly not work and will need serious programming effort before they can run again. You know how long it took for sites to upgrade from PHP 4 to 5 because of the backward compatibility issues, so imagine even more resistance and a slower uptake for PHP 6. The POSIX functions have been a core part of the language since the early days of PHP, so to suddenly dump them for no good reason will alienate and upset a large number of people. When I say "no good reason" I mean from the user's point of view. The fact that it is taking (has taken?) considerable effort to get the PCRE functions working with Unicode for PHP 6, and no-one is prepared to spend similar effort on the POSIX functions, may make it seem in the users eyes that the PHP developers are either lazy or incompetent. It does not matter that the PHP developers are unpaid volunteers - that is no reason for shoddy workmanship. If they can't do the job properly they shouldn't do it at all. BC breaks, especially on a range of functions which have been a core part of the language for many, many years, do not go down well with the users, and remember that it is the incredibly large number of people who use PHP, not the efforts of the developers, that has made it such a popular language. So annoying large numbers of users will create a PR disaster. If the PCRE functions have been made to work with Unicode, and each of the POSIX functions has a PCRE equivalent, then surely an intelligent move would be to convert all the POSIX functions into simple wrappers for their PCRE equivalents. So instead of spending huge amounts of effort in making the POSIX functions work with Unicode you simply cut out the code behind each POSIX function and replace it to a call to the relevant PCRE function. This should only require a relatively small amount of developer effort which should be balanced against the huge amount that would otherwise be required in userland. If the PHP developers are not prepared to undertake this small amount of effort then they should be prepared for considerable amounts of fallout from lots of unhappy users. I would strongly suggest that a start be made on this as soon as possible so that the changeover to PCRE wrappers can be fully tested and debugged before PHP 6 goes live. This will make the changeover from POSIX to PCRE totally transparent, and will be greatly appreciated in userland. -- Tony Marston http://www.tonymarston.net http://www.radicore.org -- PHP General Mailing List (http://www.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
