I'm looking to do my GSoC project with MariaDB, specifically the suggestion to "Port the InnoDB memcached interface to MariaDB" on the page 

https://mariadb.com/kb/en/google-summer-of-code-2014/ . 

But the project description links to a JIRA bug at 

https://mariadb.atlassian.net/browse/MDEV-4674 , 

where the author suggests that we support memcached through TCP/UDP instead, with no mention of the InnoDB plugin.

I figure that the InnoDB plugin is more important, since it's a MySQL 5.6 feature, it's widespread in industry, and the lack of a memcache interface is one of only a handful X's for MariaDB on 

http://www.skysql.com/blogs/max-mether/mysql-56-vs-mariadb-100 . 

Additionally, even without TCP/UDP support, one could just code the memcache logic manually anyway, whereas the InnoDB integration is more sophisticated and feature-rich and would be a good selling point for SkySQL.

So I'd like to start working on a proposal. Can any devs weigh in on which aspect of memcache is more desirable for MariaDB? If it sounds likes a good, straightforward GSoC project that nobody else has claimed yet?

Thanks,

Tan Tran