List:Commits« Previous MessageNext Message »
From:Sergei Golubchik Date:October 3 2008 7:33pm
Subject:Re: bzr commit into mysql-6.0 branch (marc.alff:2837) Bug#38781
View as plain text  
Hi, Marc!

On Oct 02, Marc Alff wrote:
> #At file:///home/malff/BZR-TREE/mysql-6.0-38781/
> 
>  2837 Marc Alff	2008-10-02
>       Bug#38781 (Error handling and trace logging mixed in Maria)
>       
>       --- Proof of concept patch, not ready for production ---
>       
>       This patch removes the change implemented in the server error handling
>       code (my_message_sql) for Maria, and remove the use of the following flags:
>       - ME_JUST_INFO
>       - ME_JUST_WARNING
>       
>       The design constraints applicable to the problem are:
>       - the Maria code can not have a static dependency on mysql
>       - the Maria code needs to print to the server log files (when embedded)
>       - the Maria code needs to print to stderr (when not embedded)
>       - the server needs to be independend of Maria logic in my_message_sql
>       
>       The solution consists of:
>       - defining a Plugin -> Server interface, usable by every plugin.
>         This is a general need.
>       - Provide this server interface to a plugin at initialization.
>       - For embedded code (such as unit tests or stand alone binaries like
>         maria_chk), provide an alternate implementation of this server interface.
>       
>       This patch was written as a technical proposal to resolve bug#38781,
>       and server as a base for a technical discussion.
>       It illustrates how:
>       - dependencies to the server can be abstracted in a storage engine
>       - a plugin->server interface can be implemented.

Yes, the ideas are correct. A move in the right direction, at least.

Most of that is WL#3859 (Plug-in Service API), it's being
implemented, Mats is working on it.

The rest of your patch (that isn't WL#3859) is WL#2940 (MySQL plugin
interface: error reporting) which is up for grabs. You're very welcome
to work on it (assuming your team lead agrees - but this WL is on the
6.x roadmap anyway), but then it should undergo the regular WL process,
with the design and an architecture review. A change like that cannot go
in disguised as a bugfix.

Regards / Mit vielen Grüßen,
Sergei

-- 
   __  ___     ___ ____  __
  /  |/  /_ __/ __/ __ \/ /   Sergei Golubchik <serg@stripped>
 / /|_/ / // /\ \/ /_/ / /__  Principal Software Engineer/Server Architect
/_/  /_/\_, /___/\___\_\___/  Sun Microsystems GmbH, HRB München 161028
       <___/                  Sonnenallee 1, 85551 Kirchheim-Heimstetten
Geschäftsführer: Thomas Schroeder, Wolfgang Engels, Dr. Roland Boemer
Vorsitzender des Aufsichtsrates: Martin Häring
Thread
bzr commit into mysql-6.0 branch (marc.alff:2837) Bug#38781Marc Alff2 Oct
  • Re: bzr commit into mysql-6.0 branch (marc.alff:2837) Bug#38781Konstantin Osipov3 Oct
  • Re: bzr commit into mysql-6.0 branch (marc.alff:2837) Bug#38781Sergei Golubchik3 Oct
    • Re: bzr commit into mysql-6.0 branch (marc.alff:2837) Bug#38781Konstantin Osipov3 Oct