> On Feb. 19, 2014, 8:23 p.m., Benjamin Hindman wrote: > > My hunch is that we'll actually want to inject storage into ReplicaProcess > > rather than have a factory function. This would likely be more apparent if > > we required passing 'path' to the LevelDBStorage constructor. Is there > > something that's motivating this that I'm missing? > > > > Also, s/Hided/Hid/.
The motivation is: sometimes, we want to manually inject some entries into the storage for testing (e.g. in the following large position test (https://reviews.apache.org/r/18289/)). I want to perform the manual injection at the Storage layer, rather than at the LevelDBStorage layer (because the underlying implementation might change. for example, using bdb). - Jie ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/18288/#review34917 ----------------------------------------------------------- On Feb. 19, 2014, 8:18 p.m., Jie Yu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/18288/ > ----------------------------------------------------------- > > (Updated Feb. 19, 2014, 8:18 p.m.) > > > Review request for mesos, Benjamin Hindman, Ben Mahler, and Vinod Kone. > > > Repository: mesos-git > > > Description > ------- > > See summary. Needed by a followup review on log recover test. > > > Diffs > ----- > > src/Makefile.am e02d416e5cb8edf6c44ed05c5785a5084bf02abb > src/log/replica.cpp 746d6c35c9255775ab6e70b0daf1dcecf63c16a0 > src/log/storage.hpp c0eba1b34835a89698e84c75a4c301ed7370ea1c > src/log/storage.cpp PRE-CREATION > > Diff: https://reviews.apache.org/r/18288/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Jie Yu > >
