MySQL Lists are EOL. Please join:

List:Commits« Previous MessageNext Message »
From:Chuck Bell Date:July 2 2008 2:48pm
Subject:RE: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364, Bug#35230, WL#4296
View as plain text  
Rafal,

Very well, I will make an attempt to meet your requirements. 

> But wait a moment - why are we so eager to push it now? The 
> task description is "Refine backup progress mechanism". It is 
> not fixing any serious bug. Compared to the current behaviour 
> it adds a possibility of logging to the files, not only to 
> tables. Is that functionality so important that we can not 
> release 6.0 without it? I don't think so...

I think we do have to release it in 6.0. There are some problems with the
existing mechanism which this work solves. It also introduces a fundamental
change in how logs are stored -- something that needs to be fielded by the
users so we can improve it.

> I think we should accept the fact that a task which 
> originally seemed to be simple, turned out to touch many 
> aspects of our system design which we didn't think about 
> before. Instead of hastily push a half-backed solution, we 
> should sit down and think over these aspects of the design. 

It is not a half-backed solution. It is based on the existing mechanisms and
was deliberately designed and documented in the worklog (has been for many
weeks). It was not part of this worklog to redesign any part of the existing
logging mechanisms. Please refrain from such judgments.

> Then we can use this task as a test-drive for the new design 
> solutions. We can afford that, because we have a working 
> logging mechanism which should be good enough for the first release.

I really don't like this -- it places far too much burden on us as backup
developers to improve parts of the server we don't own. I do not want this
to be another si_objects debate. We should have learned our lesson the first
time.

I will agree in part, as an interface, it should be more decoupled than this
patch is, but I think we should not redesign the MYSQL_* logging classes.
The backup logging code at that level should look the same as the existing
logging code.  

I will look at your design again and reconsider it -- so long as you
understand that this may cause planning issues and that I am doing this in
the spirit of teamwork albeit under protest. 

What disturbs me the most is we seem to have these design debates well after
the code has been written. I think there is too much emphasis on design
details during the code review. What this speaks to me more than anything is
the architecture review process is broken. 

> I can agree that all I said above doesn't apply to the 
> --backupdir option because it is a must-have feature. This is 
> yet another good reason to separate this subtask. Then I 
> would be more happy to accept an implementation even if it
>   is not fully consistent with the design ideas we have, 
> because I'd understand that we must implement --backupdir for 
> 6.0 release. Again, accepting a "temporary" patch for a task 
> which says "Implement --backupdir option" is quite different 
> from accepting such a patch for "Refine something" task.

Backupdir will be done under its bug report: BUG#35230.

Chuck

Thread
bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364, Bug#35230,WL#4296Chuck Bell25 Jun
  • Re: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364,Bug#35230, WL#4296Rafal Somla30 Jun
    • RE: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364, Bug#35230, WL#4296Chuck Bell2 Jul
      • Re: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364,Bug#35230, WL#4296Rafal Somla2 Jul
        • Re: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364,Bug#35230, WL#4296Rafal Somla2 Jul
          • RE: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364, Bug#35230, WL#4296Chuck Bell2 Jul
        • RE: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364, Bug#35230, WL#4296Chuck Bell2 Jul
  • Re: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364,Bug#35230, WL#4296Øystein Grøvlen5 Aug
    • RE: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364, Bug#35230, WL#4296Chuck Bell5 Aug
      • Re: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364,Bug#35230, WL#4296Øystein Grøvlen7 Aug
        • RE: bzr commit into mysql-6.0-backup branch (cbell:2638) Bug#33364, Bug#35230, WL#4296Chuck Bell13 Aug