incubator-cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rohit Yadav <rohit.ya...@citrix.com>
Subject Re: Tabularize the cloudmonkey response
Date Fri, 04 Jan 2013 23:42:19 GMT

On 04-Jan-2013, at 12:45 PM, Chip Childers <chip.childers@sungard.com> wrote:

> On Fri, Jan 4, 2013 at 12:18 PM, Rohit Yadav <rohit.yadav@citrix.com> wrote:
>> Yes we agreed to do cloudmonkey updates on pypi alongwith CloudStack releases, I
don't understand what you mean by syncing any updates to pypi?
>> 
>> Right now the releases I've done via pypi are not part of any acs release and I upload
a new src tarball on pypi everytime there are significant changes (if you see, it would be
once every 3-4 weeks). Also, these releases are not voted and I do them out of my interest
and free time for community, hence mentioned as community releases.
>> 
>> Pl. read my previous email following your question (syncing updates), what I asked
was:
>> 
>> a. We want to continue doing community releases on pypi to sync updates so the users
get latest and greatest changes.
>> b. We want to stop community releases and make pypi account/releases maintained by
PMC/committers so it's used to do official/voted/signed releases of cloudmonkey alongwith
ACS releases.
>> 
> 
> IMO (and I'm really looking for others to chime in and express
> opinions here too), we should really only update pypi following an
> official CloudStack release process.  To me, it's similar to the RPM /
> DEB repos that Wido maintains.  Those are distribution points for the
> project, although technically not Apache CloudStack release artifacts,
> that are updated when we do a source code release.

The only problem I see when doing cloudmonkey releases with CloudStack releases is that CloudStack
release cycles will be long, pypi releases may miss some of the useful changes which may be
frequent enough.

> 
> Then again, you (and me, because you're nice) are the one that
> controls that pypi project.  If you completely disagree, then that's
> fine.  I'm not trying to box you in or anything.  I just wanted to see
> if it would make sense to help keep the users clear about where and
> what is going on with things.  Make sense?

Meh, I don't care about the pypi, it's just a place to distribute python packages.
Any contributor, committer or PMC member who is interested, is welcome to join the pypi project
as owner or admin to maintain and release cloudmonkey via cheese shop. (unless anyone goes
wild, it's all good :)

> 
> Another idea:  we have been discussing on other threads the idea of
> adding other git repos within Apache infra.  Perhaps we should break
> Marvin and CloudMonkey out as separate repos, and allow them to follow
> a different schedule for source code releases.

I've been thinking of the same as well. With git we can have sub repos, or modules. So, you
place a git repo inside others. So, we can breakout parts of source (while retaining their
history/changes) to separate git repos, good candidates are docs (exclude apidocs), cloudmonkey,
devcloud, systemvm builder script/code, aws-api etc. This way they can be developed separately
either on asf infra or some place on github etc.

I'm inclined towards *not breaking the repo* in separate repos, but yes at least the indirect
(usage or dependency wise) pieces and *new pieces* which don't really require CloudStack can
broken out, cloudmonkey for sure qualifies.

This way CloudStack codebase can still have pointers [1] to other git repos. So, one can do
git submodule init/update etc.
Rightnow to build marvin and cloudmonkey they are required within the src as they read commands.xml
to generate cloudstackAPI python boilerplate api classes and pre cache api info for cloudmonkey.

In future after work in api_refactoring is merged, I'm planning to implement a plugin which
gives an api to load available api info for a user (apiname, endpoints, descp, help docs,
param info etc.) from the mgmt server on the fly (hence no preaching required). If this works,
changes in apis won't affect cloudmonkey or any client which implements in a generic way (functional
programming ftw!) and clients (cloudmonkey, marvin) can be extracted out of CloudStack and
be developed as separate projects.

[1] http://git-scm.com/book/en/Git-Tools-Submodules

> 
> thoughts?
> 
>> Regards.
>> 
>> 
>> ________________________________________
>> From: Chip Childers [chip.childers@sungard.com]
>> Sent: Friday, January 04, 2013 8:04 PM
>> To: cloudstack-dev@incubator.apache.org
>> Subject: Re: Tabularize the cloudmonkey response
>> 
>> On Thu, Jan 3, 2013 at 5:23 PM, Rohit Yadav <rohit.yadav@citrix.com> wrote:
>>> Sure, the pypi community release 4.0.0-2 (http://pypi.python.org/packages/source/c/cloudmonkey/cloudmonkey-4.0.0-2.tar.gz#md5=6125f804bc1dec4447df1e1b48a7d85b)
has the latest changes with tabular output etc. I can upload the latest tar ball that would
make it 4.0.0-3?
>>> 
>>> Or, for project releases, should we do a 4.0.1 release for cloudmonkey when CloudStack
releases after voting etc. or should we go ahead with the community releases (4.0.0-x, cloudmonkey
versions are like this: major cloudstack version hyphen last cloudmonkey revision+1) before
the actual release?
>>> 
>> 
>> I thought we agreed that cloudmonkey updates to pypi would be part of
>> the overall cloudstack release process.  Specifically, it's only
>> updated when we release a new source release.  This would align with
>> the DEB and RPM package release process, and we may end up doing
>> something similar with maven artifacts as well.
>> 
>>> Regards.
>>> 
>>> On 03-Jan-2013, at 12:38 PM, Chip Childers <chip.childers@sungard.com>
wrote:
>>> 
>>>> Hey Rohit,
>>>> 
>>>> I think that we should probably be syncing any updates to cloudmonkey
>>>> on pypi with project releases.  What do you think?
>>>> 
>>>> On Thu, Dec 20, 2012 at 8:00 PM, Rohit Yadav <rohit.yadav@citrix.com>
wrote:
>>>>> If you want to have the tabularized output, upgrade using: pip install
--upgrade cloudmonkey
>>>>> Updated wiki, see tabular output usage on wiki:
>>>>> https://cwiki.apache.org/confluence/display/CLOUDSTACK/CloudStack+cloudmonkey+CLI
>>>>> 
>>>>> Regards.
>>>>> 
>>>>> On 20-Dec-2012, at 4:37 PM, "Rohit Yadav (JIRA)" <jira@apache.org>
wrote:
>>>>> 
>>>>>> 
>>>>>>   [ https://issues.apache.org/jira/browse/CLOUDSTACK-545?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
>>>>>> 
>>>>>> Rohit Yadav resolved CLOUDSTACK-545.
>>>>>> ------------------------------------
>>>>>> 
>>>>>>  Resolution: Fixed
>>>>>>    Assignee: Rohit Yadav
>>>>>> 
>>>>>> Fixed on master:
>>>>>> 
>>>>>> CLOUDSTACK-545: Tabularize cloudmonkey's response using filter=<fields,>
argument
>>>>>> 
>>>>>> authorRohit Yadav <bhaisaab@apache.org>
>>>>>> Fri, 21 Dec 2012 00:28:32 +0000 (16:28 -0800)
>>>>>> committerRohit Yadav <bhaisaab@apache.org>
>>>>>> Fri, 21 Dec 2012 00:33:08 +0000 (16:33 -0800)
>>>>>> commit8f51c630bdb874f9522ec3873c881b1c4e7cfdf7
>>>>>> 
>>>>>>> Tabularize the cloudmonkey response
>>>>>>> -----------------------------------
>>>>>>> 
>>>>>>>              Key: CLOUDSTACK-545
>>>>>>>              URL: https://issues.apache.org/jira/browse/CLOUDSTACK-545
>>>>>>>          Project: CloudStack
>>>>>>>       Issue Type: Bug
>>>>>>>   Security Level: Public(Anyone can view this level - this is
the default.)
>>>>>>>       Components: Cloudmonkey, Test Tools
>>>>>>>         Reporter: Prasanna Santhanam
>>>>>>>         Assignee: Rohit Yadav
>>>>>>>      Attachments: 0001-cli-cloudmonkey-tabular-output-and-result-view-filte.patch
>>>>>>> 
>>>>>>> 
>>>>>>> cloudmonkey responses would be more sensible and easier on the
eye if they can look tabular similar to the ec2 tools. enhance cloudmonkey to support tabular
formatting instead of sequential lists of the response
>>>>>> 
>>>>>> --
>>>>>> This message is automatically generated by JIRA.
>>>>>> If you think it was sent incorrectly, please contact your JIRA administrators
>>>>>> For more information on JIRA, see: http://www.atlassian.com/software/jira
>>>>> 
>>>>> 
>>> 
>>> 
>> 


Mime
View raw message