Posted on 2013-12-23 by Benjamin Eberlei
Facebook has been pushing HHVM alot lately , helping open source projects to get their test-suite running 100%. For Doctrine HHVM is particularly interesting, because of the performance gains that the complex PHP algorithms inside ORM would probably get. From my current feeling Doctrine will be the PHP open-source project getting the most gain from running on HHVM. However with the tests not yet passing on the ORM, we can only imagine how big that performance improvement will be.
One roadblock for us to investigate HHVM in more detail was missing CI support. But then Travis CI announced support for HHVM last week. With automated testing support available we think it is time to announce our official HHVM roadmap.
One of our goals for 2014 is running DBAL and ORM on HHVM with 100% of
the testsuites passing. Every Doctrine subproject targeting to support
HHVM will start running the tests against HHVM with
enabled on Travis CI. Whenever a Doctrine subproject passes all its
tests on HHVM, we will remove the
allow_failure and the project will
be officially supporting HHVM from that version on.
So far we have been working on the Common projects to run on HHVM for several months now and Guilherme and Alexander are contributing to HHVM itself to get some missing APIs working. We are happy to announce that the following Common projects currently have full HHVM support from us:
Guilherme is working on getting Annotations and Cache working and the Common mainproject will be evaluated shortly after all the common projects succeed. DBAL and ORM will be much more work, but we are very confident to achieve this goal.
If you want to help us with this goal, you can check the current Travis failure reports of the projects and come up with ideas how to fix them in the Doctrine code or with bug reports for HHVM. We are glad to discuss these issues on Freenode IRC in channel "#doctrine-dev".
With this announcement we hope that other PHP projects, frameworks and libraries will follow to make HHVM an official build target in the future.