I am a strong supporter of the non redundancy rule on PEAR. That said I am voting +1 because:
- Savant has a totally different approach then XIPE,Flexy and any other template engine available on PEAR. It is lightweight not compiling, non caching and it is exremely easy to extend it using filters.
There are different ways to look at it but if you try both (as I did) the other existing template engines you will see the difference from a developer standpoint.
Even if there is a subtle line on defining redundant packages, Savant didn't cross it.
In fact Savant is not an engine but an OO template system. This alone is a proof that is not a duplicate of an existing package in PEAR.
-Savant is already PEAR-compliant and well documented. The documentation issue is often treated as secondary but is not. If we want PEAR to stay as a quality repository, poorly documented package should really shape up.
The success of PHP was also contributed by the excellent online documentation. The same applies to PEAR. Savant fits the bill greatly.
-Templates are not like every other package. If a developer decides to adopt a template engine it will be using it daily. This category cannot be seen like Networking for instance. If you do have a Net_Curl or Net_FTP it would really make no sense to have 2 packages doing the same thing even with a different structure.
For this very reason offering a reasonable numbers of template engines/systems with marked differences will be on the own interest of PEAR.
Eventually I would be strongly against any integration with Flexy. I wouldn't use Savant as a sub method of another package. It's different and very lightweight.
Embedding it into something else would be like embedding sqlite within MySQL.
To summarize: It's different, well documented, well coded and will provide a benefit to PEAR and its users.