Hello Gregory, the good news on this fron is that re2c meanwhile has a new beta version with flex like conditoin support. Also when changing from flex to re2c we should switch from bison to lemon. All in all a damn huge amount of work. But the result would be much cleaner, faster, more flexible, fully reentrant, reuseable outside of the engine and cabable of unicode.... Yet you mentioned the big issue, someone would need to do this. Though the biggest entry on the to do would be discussing it ... hopefully not needting to rename something
marcus Friday, August 17, 2007, 7:35:43 PM, you wrote: > Ralph Schindler wrote: >> Hey David & Lucas, >> I brought this up a year ago on the list here: >> >> http://marc.info/?t=114728193700003&r=1&w=2 >> >> Essentially, the trend will continue to be that as keyword requiring >> features are added, the global naming space will continue to become >> smaller and less safe for older applications written for older versions >> of php. >> >> Until it becomes a goal of php-internals to add contextual awareness and >> sensitivity to the compiler's tokenizer, this will always be a problem. >> >> Here is a more amusing list of method names that will also not work: >> http://marc.info/?l=php-dev&m=114736739115429&w=2 >> >> And the general opinion summed up by Wez ;) >> >> http://marc.info/?l=php-dev&m=114764099400409&w=2 > Hi, > Doing this with flex is non-trivial. It's a lot easier with a > re2c-based lexer, but porting the flex scanner to re2c is the definition > of "oh-my-god-almighty" complex and until someone just does it, it ain't > gonna happen. > The other hackish alternative is to add rules to the parser with > reserved words (T_FUNCTION T_IMPORT as well as T_FUNCTION T_STRING, etc. > etc.), but this slows down the parser and adds unnecessary duplication. > Greg Best regards, Marcus -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php
