The problem I see with template_compile template mods is that, unless the user is running Template Modification System, you would essentially be forcing the user's database to customize that template in each style. Once this happens, it is difficult to determine whether the user actually has other customizations in the template or not, at least not without an expensive 3-way comparison that uses many megabytes of memory. For users that don't have the template merger class (pre-vB4 users) or that are on shared hosting, we can't revert the changes reliably during a product uninstall, upgrade, or discovery of TMS since the last update.
For these reasons I think it is less intrusive for the user and less error prone if we just encourage users to use TMS, and if TMS is not present, to use the inefficient runtime cache edit.
|