Is redis orphaned?

Jon jdisnard at gmail.com
Wed Jun 18 17:22:31 UTC 2014


On Wed, Jun 18, 2014 at 10:41 AM, Christopher Meng <cickumqt at gmail.com> wrote:
> I'm asking the original owner of redis to see why he chose tcmalloc as
> memory allocator, as I couldn't find any strong opinion to introduce
> gperf dependency just because of some negligible(not sure, though)
> performance improvement. Moreover, jemalloc is the default option for
> redis on Linux.
>

Unless there is a really great reason to use tcmalloc, I would say the
default jemalloc should happen.
My experience with the two is JEmalloc has less issues long term
bloating memory heaps, with comparable performance.
But my reason to support JEmalloc is purely a releng driven thing,
reducing BR's and R's is always preferable.

My two-bits, but it would be nice to have a test suite to measure the two.
But that requires effort and time, which are always low supply.
I'd say give the maintainer time to chime in, otherwise make it happen
after reasonable time-out.


-- 

-Jon Disnard


More information about the devel mailing list