Re: [Sqlgrey-users] 1.6.x and 1.7.x freezes

[Lionel B. <lio...@bo...>]

Rupa Schomaker (lists) wrote the following on 26.07.2005 18:27 :

>On 7/26/2005 4:27 AM, Lionel Bouton wrote:
>  
>
>>>Is it clean_method or cleanup_method?  I now have both in my
>>>sqlgrey.conf file set to sync.
>>>
>>>75: $dflt{clean_method}       = 'sync';
>>>and then
>>>1403:    if ($self->{sqlgrey}{cleanup_method} eq 'sync') {
>>>      
>>>
>>Hum...
>>
>>Good catch. I'll patch the code to properly support "clean_method" as
>>specified in the example sqlgrey.conf. Putting "cleanup_method = sync"
>>in sqlgrey.conf should indeed avoid the fork in 1.6.4
>>    
>>
>
>It also looks like not all places use start_cleanup() and instead call
>fork_cleanup() instead.  Also, the initialization of the sqlgrey hash
>doesn't set cleanup_method which resulted in async being used no matter
>what.
>
>Patch attached.
>
>  
>

Good catch again :-)
This start/fork_cleanup mixup is odd, I remember having checked an 
intermediate version for such cases (I was initialy unsure on how to 
name these methods) and corrected them. I may have mixed up borked and 
cleaned version at some point before commiting...

I'll drop support for cleanup_method and only leave support for 
clean_method (TODO: make the config parser aware of supported options 
and complain when unknown are found). This was the only config entry 
advertised in the etc/sqlgrey.conf example.

Preparing a 1.6.5 version now.

Lionel

PS: your strace following the two processes might help understand what's 
going on when using tha async method, I'll try to give it a shot later.