maria-developers team mailing list archive

["Vladislav Vaintroub" <wlad@xxxxxxxxxxxxxxxx>]

> -----Original Message-----
> From: maria-developers-
> bounces+wlad=montyprogram.com@xxxxxxxxxxxxxxxxxxx [mailto:maria-
> developers-bounces+wlad=montyprogram.com@xxxxxxxxxxxxxxxxxxx] On Behalf
> Of Daniel Bartholomew
> Sent: Donnerstag, 15. März 2012 20:30
> To: Maria Developers; Maria Docs
> Subject: [Maria-developers] MariaDB 5.5.21
> 
> All,
> 
> I've uploaded the MariaDB 5.5.21-beta packages to the primary mirror
> and within a few hours the files will be distributed to all of the
> mirrors.
> 
> Over the past couple of weeks I've added several items to the release
> notes page for 5.5.21 and today I created the changelog page. The pages
> are here:
> 
> http://kb.askmonty.org/en/mariadb-5521-release-notes
> http://kb.askmonty.org/en/mariadb-5521-changelog
> 
> Once the files are distributed I will update the pages and add
> in the download links and remove the "unreleased" banners from the
> pages linked to from the release notes page. I will also send out an
> announcement email, and post announcements to twitter, FB, G+, and so
> on as usual.
> 
> Please let me know ASAP if there is anything which is not on the
> release notes page which should be (or jump in and add it).
Hi Daniel,
this is the first release to feature the new threadpool.
- link to the documentation :
http://kb.askmonty.org/en/thread-pool-in-mariadb-55 
- link to preliminary benchmarks :
http://kb.askmonty.org/en/threadpool-benchmarks . NOTE: Axel ran his own
benchmarks recently too, and will probably blog/publish them soon. The
throughput curves  will be similar as far as I can tell.

- I added some time ago to "What is new in 5.5" 
  Significantly improved for efficiency new thread pool , comparable in
functionality to the closed source feature in MySQL Enterprise

NOTE:  pool-of-threads is now default scheduler on Windows Vista+ , so this
is a change our users should be aware of . I did not made it default on
Unixes (even if we had  good benchmarks) yet, because Linux users are picky,
and Monty objected to do it just yet , and it had more own coding than
Windows case(where we rely on OS threadpooling functionality). So Unix
implementation should probably be better tested for corner cases in the
field, before it could be declared overall much better.

Can you add essence of the above to the release notes?

Thanks,
Wlad