← Back to team overview

randgen team mailing list archive

Re: MariaDB patches


Hi Roel, Bernt, all,

@all: I'll wait for another day for further remarks, since I initially promised to wait for two days.

@Roel, Bernt: comments inline.

On 6/19/2013 5:27 PM, Roel Van de Paar wrote:
> 856: Print PID in test output
> Please make this one optional behavior.

I just won't merge it.

On 6/20/2013 1:46 AM, Bernt Marius Johnsen wrote:

On 06/19/2013 01:42 AM, Elena Stepanova wrote:

874: Minimize the number of non-default startup options

Although I agree with this fix in principle, I'm hesitant to accept it
since it affects a vast number of existing tests.

OK, will not be merged, your choice.

877: There is no need to override the previous error log -

I disagree (but I don't scream, I never do ;-) ). Collecting error log
files from several runs should be the task of the surrounding framework.
Remember that most of the RQG runs are automated in other frameworks (or
at least we do that to a great degree now).


(For a note, it's not about several runs. The surrounding code already takes care of it: the whole vardir is deleted in the same module, by createMysqlBase subroutine. So whenever you call MySQLd->new before server startup, there will be no old error log anyway. The change could only have effect when you call MySQLd->startServer() *without* calling createMysqlBase() first -- that is, when you restart the server at runtime. But you don't, at least not in the randgen public tree; and if you did, you would want the log to be either continuous or rotated, but not overridden).

But your wish is my command :)
Will not be merged.

All rejected changes will still be in my tree, as well as whatever else I add or change from this point and till the next merge, so if anybody wants them, you can find them there.


Follow ups