> On Dec. 18, 2014, 6:05 p.m., Adam B wrote:
> > include/mesos/resources.hpp, lines 64-69
> > <https://reviews.apache.org/r/28698/diff/2/?file=793361#file793361line64>
> >
> >     Does this always assume STATIC? Shouldn't you be able to parse a string 
> > into name/value/role while also specifying the reservation type?
> 
> Michael Park wrote:
>     This one was intentionally put off for 2 reasons.
>       1. We would have to change the syntax for resources. Current syntax is 
> "name(role):value;...", how should we introduce the dynamic reservation 
> there? perhaps "name[role]:value"? I mentioned this to Tim Chen and he 
> suggested maybe we just ditch this and resort to a JSON format or something 
> more explicit.
>       2. I don't think we quite want to allow slave resources on the command 
> line to include dynamic reservations. If we allow dynamic reservations to be 
> specified here, we would have to explicitly check for this in the command 
> line and maybe other places we want to disallow it.

Fair enough. So how would it be checkpointed and recovered? JSON 
ser/deserialization of the protobuf?


> On Dec. 18, 2014, 6:05 p.m., Adam B wrote:
> > src/common/resources.cpp, line 492
> > <https://reviews.apache.org/r/28698/diff/2/?file=793362#file793362line492>
> >
> >     Is CHECK appropriate here, or should we instead return a Try<Resources>?
> 
> Michael Park wrote:
>     I think returning a `Try<Resources>` may be better, but I think that 
> probably breaks frameworks. What's our preferred solution in these situations?

I doubt anybody's actually using flatten in a framework. But if you want to do 
due diligence, you could send a short warning/announcement email to the dev@ 
mailing list, and maybe post a small blurb in upgrades.md.


- Adam


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/28698/#review65561
-----------------------------------------------------------


On Dec. 30, 2014, 4:04 p.m., Michael Park wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/28698/
> -----------------------------------------------------------
> 
> (Updated Dec. 30, 2014, 4:04 p.m.)
> 
> 
> Review request for mesos, Adam B, Benjamin Hindman, Ben Mahler, Jie Yu, and 
> Vinod Kone.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> Modified Resources to account for reservation type.
> 
> 
> Diffs
> -----
> 
>   include/mesos/resources.hpp f1517b73def9aff4039f95a89f66208ba1d21c0f 
>   src/common/resources.cpp c17e1791130e7d545bb7cdd54d97d65325d3a69e 
> 
> Diff: https://reviews.apache.org/r/28698/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Michael Park
> 
>

Reply via email to