rockandroller

These are very good points, thank you. I guess this is the main drawback of not wanting to 'reinvent the wheel' - building upon existing CMS's and frameworks means I am choosing to rely on those people to 'do the right thing' with regards to 'the future'...

* I can't quite see the forum where this question exactly belongs, so moderators, please relocate as appropriate... I know a lot of PHP development is TEMPORARY and not intended to last forever ( I used to code for an advertising company). However, there are some things I would like to see endure... and I was really caught short a couple years back when my hosting provider (godaddy) KILLED PHP4 altogether. I had a couple of sites that PERISHED because they were built on frameworks that were using PHP4, and could't reasonably be converted to PHP5. Migration requirements were simply too pervasive in our case, would have required many many hours of rebuilding functions I know I had the "option" to upgrade from shared hosting to a dedicated server ( and install my own PHP4 on it!) - but I simply couldn't afford a tenfold jump in hosting costs. Now, the same hosting provider is only officially supporting PHP version 5.3 - with 5.2 as "legacy" language. It seems that the days for 5.2 will be numbered ( and who knows - probably 5.3 down the road a bit, as well!) I notice the 5.2 to 5.3 migration guide shows a dozen lines of "backwards incompatible" changes - and the same for 5.3 to 5.4 ... I rely on various frameworks and CMS's to get things done, and cringe at the thought that some day soon I could be faced with 8 zillion lines of code that need massaging. How does everyone else look at this situation? I need to decide whether to extend, or toss and start over some existing projects. I certainly understand the value of the latest supported language versions, but would hate to have the same problem sneak up and bite me 2 or 3 years down the road! I can't seem to find a really good "PHP language version roadmap" to make timeline decisions... not on PHP.net, anyway. And the wikipedia page does not look too promising...