> On April 17, 2016, 11:47 p.m., Jiang Yan Xu wrote: > > include/mesos/mesos.proto, line 762 > > <https://reviews.apache.org/r/45958/diff/1/?file=1337790#file1337790line762> > > > > How about `optional SharedInfo shared`? > > > > [shareable](http://www.merriam-webster.com/dictionary/shareable): > > capable of being shared > > > > The following sentences read pretty clear to me. > > - Some resource types in Mesos are shareable. > > - Currently only persistent volumes are shareable. (This has nothing to > > do with whether `SHARE` operation has been applied, just that this type of > > resource can be made shared.) > > - The `SHARE` operation makes a nonshared resource **shared**. > > - The `UNSHARE` operation makes **shared** resource nonshared. > > - `SharedInfo` is currently empty but in the future we may add policies > > around how this resource should be **shared**. > > > > Plus we can compare this with `shared_ptr` which is semantically very > > similar. > > > > If we agree to this please also change the use of these words elsewhere > > appropriately. > > Anindya Sinha wrote: > I think ShareInfo seems fine to me. However, I think if there is a strong > opinion regarding this, I think Shareable is better simply because it > describes the resource (ie. adjective) and is on the same principles as > Revocable (as pointed by Guangya Liu). > However, I stringly believe ShareInfo should be fine. > > Guangya Liu wrote: > In my understaind, I think that this is similar with `RevocableInfo` as > following: > > message RevocableInfo {} > > // If this is set, the resources are revocable, i.e., any tasks or > // executors launched using these resources could get preempted or > // throttled at any time. This could be used by frameworks to run > // best effort tasks that do not need strict uptime or performance > // guarantees. Note that if this is set, 'disk' or 'reservation' > // cannot be set. > optional RevocableInfo revocable = 9; > > Agree with Anindya, using the concept of `Shareable` will have same > principal with `Revocable`.
`shared` and `shareable` are both adjectives, `share` is a verb. `shareable` and `shared` have similar meanings except that former emphasizes on **capability** and the latter on the **state**: operation SHARE marks a resource shared just like RESERVE marks a resource reserved (`Resources::isReserved()`), even though in protobuf it's captured by a noun `reservation`. Besides IMO **shared** describes the **status** better, my concerns is that in the future we'll need APIs for determining if some resources **are capable of being shared** (`Resources::isShareable(...)`) and we lose the word to conveniently describe it if we use it for something else now. (If this was a one off thing we wouldn't even call it shared resources, just shared volumes). My apologies that dicussions around this should have been captured earlier in the design phase but since it's a public API I think we should be more rigorous otherwise it requires deprecation and carefully orchestrated upgrades, etc. to change it in the future. Thoughts? - Jiang Yan ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/45958/#review128548 ----------------------------------------------------------- On April 8, 2016, 4:16 p.m., Anindya Sinha wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/45958/ > ----------------------------------------------------------- > > (Updated April 8, 2016, 4:16 p.m.) > > > Review request for mesos and Jiang Yan Xu. > > > Bugs: MESOS-4892 > https://issues.apache.org/jira/browse/MESOS-4892 > > > Repository: mesos > > > Description > ------- > > Added ShareInfo in Resource protobuf to allow for sharing of resources. > > > Diffs > ----- > > include/mesos/mesos.proto 63c181ae0a1e350fc27e36b1698e02db100b8861 > include/mesos/v1/mesos.proto a60a834e2538d54db7f257a0d4adfbb503ec1b0f > > Diff: https://reviews.apache.org/r/45958/diff/ > > > Testing > ------- > > Tests successful. > > > Thanks, > > Anindya Sinha > >
