On 07/23/10 12:59 AM, Keith Mitchell wrote:
> Couple quick nits...
>
> On 07/22/10 04:33 PM, Alok Aggarwal wrote:
>> The most recent version of the DC design doc specifies
>> user defined checkpoints as follows:
>>
>> <execution stop_on_error="false">
>> <checkpoint name="custom-script" mod_path="custom_script"
>> checkpoint_class="CustomScript">
>> <args>/usr/bin/myscript.sh arg1 arg2</args>
>> </checkpoint>
>> </execution>
>>
>> As was pointed out in this morning's meeting, this has
>> the downside of not allowing for a way to specify arguments
>> to myscript.sh that get reside in the DOC.
>>
>> So, I'm enhancing my proposal based on some conversations
>> with Darren and Keith.
>>
>> For arguments that need to be referenced from the DOC, allow
>> XPATH style argument specification. So, if the 'distro_name'
>> element needs to be referenced from the DOC, the args to
>> myscript.sh would be specified as -
>>
>> %{/doc/volatile/dc_spec/distro_name}
>>
>> Or, simply,
>>
>> %{/*/distro_name}
>
> Perhaps we should avoid the asterisk as its a special character.
>
> %{//distro_name} (double slash instead of single slash)
> The above format is more closely similar to xpath style, as well.
That works for me.
>
>>
>> Anytime, the CustomScript checkpoint sees a %{ .. }, it will perform a
>> substitution based on a call to the DOC.
>> So, the above two specifications will respectively map to:
>>
>> doc.volatile.get_children(name=distro_name)
>>
>> and
>>
>> doc.get_descendents(name=distro_name)
>>
>> If a user wants to reference a more specific attribute, it
>> can be specified as follows -
>>
>> %{/doc/volatile/n...@attribute}
>>
>> and this will get substituted to the appropriate DOC call
>> and object.__dict__['attribute'] referenced.
>
> When you get to implementation time, you'll want to use "getattr"
> instead of going directly to the __dict__ object.
Makes sense.
>
>>
>> If multiple objects match a given path, a subscript reference
>> can be specified to reference a specific entry.
>>
>> Does this seem workable?
>
> Sure does!
I'll take a look at this from the DOC point of view then...
Thanks,
Darren.
>
> Thanks,
> Keith
>
>>
>> Alok
>> _______________________________________________
>> caiman-discuss mailing list
>> [email protected]
>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>
> _______________________________________________
> caiman-discuss mailing list
> [email protected]
> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
