ourdelta-developers team mailing list archive

[Arjen Lentz <arjen@xxxxxxxxxxxxxxxx>]

Hi all,

We'll keep looking at some more 5.0 stuff, but now that the existing  
patches are clean and sorted, we can look at porting them to 5.1.
This will involve some manual labour - not a huge amount of server- 
guru-foo though, common sense should guide you most of the way.
If you want to tackle a patch, just branch off trunk, copy a patch  
across from /mysql/5.0 to /mysql/5.1, and add it to the 5.1 patch list.
At this stage, feel free to pick any, we can sort out the dependencies  
fairly easily still.

You can, at first, just try a quilt push to see if the 5.0 patch still  
applies. It might fail on some files.
If you use --force and/or --keep-rejects then you will be able to  
review what's going on.
In some cases, the shortest method might just be to load up the raw  
patch in one window, 5.1 source file in another, and work it out  
manually.
The way I've refreshed patches with quilt means that the @@ headers of  
each patch section contain the function name where the patch lines are  
applied, so you have that information on top of the usual +/- 3 lines  
context. That should help!

If you get stuck with anything, just send a message here, there's many  
skilled people around who might not have time right now for hacking  
themselves, but will certainly be happy to help things along by  
answering your question.


You may wonder, what about the 5.1 Percona patches. Well, a few of  
those are fine, however others are different from their identically  
named 5.0 version, or we've modified the patch in 5.0. As a practical  
example, the userstats patch in 5.0 is V2 from Google, while in 5.1  
it's V1 which also doesn't support information_schema. You can always  
do a compare (also check the version in the patch_info part of the  
patch) to see what might be different. Or ask! I've tried to document  
most things I've found over the last few weeks, but some things may  
still just be in my head.

Oh, about information schema. In 5.1, we'll want to make any I_S  
addition into a plugin, that way we steer clear of the nasty parser  
patch conflicts that we now have in 5.0 (patch lines within  
eachother's context, making patches interdependent). But you don't  
have to worry about that either, if you can make a patch -or part of  
it- work in 5.1, we will sort out the I_S plugin foo a bit later.


It'd be great to get our 5.1 patchset sorted within say 3 weeks, with  
build time that puts us in line with the GA release of 5.1.
We may also be able to do some extra work on this at the OpenSQLCamp  
next week.

Cheers,
Arjen.
--
Arjen Lentz, Director @ Open Query (http://openquery.com.au)
Training and Expertise for MySQL in Australia and New Zealand

OurDelta: free enhanced builds for MySQL @ http://ourdelta.org