Re: [openstack-dev] [mistral][osc-lib][openstackclient] is it too early for orc-lib?

[Hardik]

Hi,

Regarding osc-lib we have mainly two changes.

1) Used "utils" which is moved from openstackclient.common.utils to 
osc_lib.utils

2) We used "command"  which wrapped in osc_lib from cliff.

So I think there is no harm in keeping osc_lib.

Also, I guess we do not need openstackclient to be installed  with 
mistralclient as if mistral is used in standalone mode

there is no need of openstackclient.

Thoughts ?

Thanks and Regards,
Hardik Parekh


On Thursday 30 June 2016 05:25 PM, Renat Akhmerov wrote:

Hi,

We already let osc-lib in to Mistral but I found out that such 
transition was blocked in TripleO, [1].
I'd like to ask the team to read into it and discuss whether we need 
to revert corresponding patches in Mistral or not.


[1] https://review.openstack.org/#/c/11/

Renat Akhmerov
@Nokia



__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [mistral][osc-lib][openstackclient] is it too early for orc-lib?

[Dean Troyer]

On Thu, Jun 30, 2016 at 8:38 AM, Hardik 
wrote:

> Regarding osc-lib we have mainly two changes.
>
> 1) Used "utils" which is moved from openstackclient.common.utils to
> osc_lib.utils
> 2) We used "command"  which wrapped in osc_lib from cliff.
>
> So I think there is no harm in keeping osc_lib.
>

Admittedly the change to include osc-lib is a little early, I would have
preferred until the other parts of it were a bit more stable.


> Also, I guess we do not need openstackclient to be installed  with
> mistralclient as if mistral is used in standalone mode
> there is no need of openstackclient.
>

The choice to include OSC as a dependency of a plugin/library rests
entirely on the plugin team, and that will usually be determined by the
answer to the question "Do you want all users of your library to have OSc
installed even if they do not use it?"  or alternatively "Do you want to
make your users remember to install OSC after installing the plugin?"

Note that we do intend to have the capability on osc-lib to build an
OSC-like stand-alone binary for plugins that would theoretically make
installing OSC optional for stand-alone client users.  This is not complete
yet, and as I said above, one reason I wish osc-lib had not been merged
into plugin requirements yet.  That said, as long as you don't use those
bits yet you will be fine, the utils, command, etc bits are stable, it is
the clientmanager and shell parts that are still being developed.

dt

-- 

Dean Troyer
dtro...@gmail.com
__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [mistral][osc-lib][openstackclient] is it too early for orc-lib?

[Steve Martinelli]

The crux of this, as Dean stated, is if the library wants OSC to always be
pulled in (along with its many dependencies). We've seen folks include it
in requirements, test-requirements, or even not at all (just document that
OSC needs to be installed).

I tossed up the idea with the ironic team of leveraging "extras" field to
list OSC as optional, the change would look like:

--- a/setup.cfg
+++ b/setup.cfg
@@ -22,6 +22,10 @@ classifier =

+[extras]
+cli =
+  python-openstackclient>=3.0.0  # Apache-2.0
+

So, if a user wanted to install just the python binding of ironicclient or
mistralclient, they would do $ pip install python-ironicclient; if a user
wanted the CLI as well.. $ pip install python-ironicclient[cli]

Just an idea, it may be overkill and completely horrible.

On Thu, Jun 30, 2016 at 5:29 PM, Dean Troyer  wrote:

> On Thu, Jun 30, 2016 at 8:38 AM, Hardik 
> wrote:
>
>> Regarding osc-lib we have mainly two changes.
>>
>> 1) Used "utils" which is moved from openstackclient.common.utils to
>> osc_lib.utils
>> 2) We used "command"  which wrapped in osc_lib from cliff.
>>
>> So I think there is no harm in keeping osc_lib.
>>
>
> Admittedly the change to include osc-lib is a little early, I would have
> preferred until the other parts of it were a bit more stable.
>
>
>> Also, I guess we do not need openstackclient to be installed  with
>> mistralclient as if mistral is used in standalone mode
>> there is no need of openstackclient.
>>
>
> The choice to include OSC as a dependency of a plugin/library rests
> entirely on the plugin team, and that will usually be determined by the
> answer to the question "Do you want all users of your library to have OSc
> installed even if they do not use it?"  or alternatively "Do you want to
> make your users remember to install OSC after installing the plugin?"
>
> Note that we do intend to have the capability on osc-lib to build an
> OSC-like stand-alone binary for plugins that would theoretically make
> installing OSC optional for stand-alone client users.  This is not complete
> yet, and as I said above, one reason I wish osc-lib had not been merged
> into plugin requirements yet.  That said, as long as you don't use those
> bits yet you will be fine, the utils, command, etc bits are stable, it is
> the clientmanager and shell parts that are still being developed.
>
> dt
>
> --
>
> Dean Troyer
> dtro...@gmail.com
>
> __
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [mistral][osc-lib][openstackclient] is it too early for orc-lib?

[Renat Akhmerov]

Ok, based on what has been said here I suggest we keep this code for now. The 
changes were really minimal. If it creates some problems for us we can always 
easily revert.

Renat Akhmerov
@Nokia

> On 01 Jul 2016, at 04:57, Steve Martinelli  wrote:
> 
> The crux of this, as Dean stated, is if the library wants OSC to always be 
> pulled in (along with its many dependencies). We've seen folks include it in 
> requirements, test-requirements, or even not at all (just document that OSC 
> needs to be installed). 
> 
> I tossed up the idea with the ironic team of leveraging "extras" field to 
> list OSC as optional, the change would look like:
> 
> --- a/setup.cfg
> +++ b/setup.cfg
> @@ -22,6 +22,10 @@ classifier =
>  
> +[extras]
> +cli =
> +  python-openstackclient>=3.0.0  # Apache-2.0
> +
> 
> So, if a user wanted to install just the python binding of ironicclient or 
> mistralclient, they would do $ pip install python-ironicclient; if a user 
> wanted the CLI as well.. $ pip install python-ironicclient[cli] 
> 
> Just an idea, it may be overkill and completely horrible.
> 
> On Thu, Jun 30, 2016 at 5:29 PM, Dean Troyer  > wrote:
> On Thu, Jun 30, 2016 at 8:38 AM, Hardik  > wrote:
> Regarding osc-lib we have mainly two changes.
> 
> 1) Used "utils" which is moved from openstackclient.common.utils to 
> osc_lib.utils
> 2) We used "command"  which wrapped in osc_lib from cliff.
> 
> So I think there is no harm in keeping osc_lib.
> 
> Admittedly the change to include osc-lib is a little early, I would have 
> preferred until the other parts of it were a bit more stable.
>  
> Also, I guess we do not need openstackclient to be installed  with 
> mistralclient as if mistral is used in standalone mode
> there is no need of openstackclient.
> 
> The choice to include OSC as a dependency of a plugin/library rests entirely 
> on the plugin team, and that will usually be determined by the answer to the 
> question "Do you want all users of your library to have OSc installed even if 
> they do not use it?"  or alternatively "Do you want to make your users 
> remember to install OSC after installing the plugin?"
> 
> Note that we do intend to have the capability on osc-lib to build an OSC-like 
> stand-alone binary for plugins that would theoretically make installing OSC 
> optional for stand-alone client users.  This is not complete yet, and as I 
> said above, one reason I wish osc-lib had not been merged into plugin 
> requirements yet.  That said, as long as you don't use those bits yet you 
> will be fine, the utils, command, etc bits are stable, it is the 
> clientmanager and shell parts that are still being developed.
> 
> dt
> 
> -- 
> 
> Dean Troyer
> dtro...@gmail.com 
> 
> __
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe 
> 
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev 
> 
> 
> 
> __
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [mistral][osc-lib][openstackclient] is it too early for orc-lib?

[Dmitry Tantsur]

On 06/30/2016 11:29 PM, Dean Troyer wrote:

On Thu, Jun 30, 2016 at 8:38 AM, Hardik
mailto:hardik.par...@nectechnologies.in>> wrote:

Regarding osc-lib we have mainly two changes.

1) Used "utils" which is moved from openstackclient.common.utils to
osc_lib.utils
2) We used "command"  which wrapped in osc_lib from cliff.

So I think there is no harm in keeping osc_lib.


Admittedly the change to include osc-lib is a little early, I would have
preferred until the other parts of it were a bit more stable.


Also, I guess we do not need openstackclient to be installed  with
mistralclient as if mistral is used in standalone mode
there is no need of openstackclient.


The choice to include OSC as a dependency of a plugin/library rests
entirely on the plugin team, and that will usually be determined by the
answer to the question "Do you want all users of your library to have
OSc installed even if they do not use it?"  or alternatively "Do you
want to make your users remember to install OSC after installing the
plugin?"

Note that we do intend to have the capability on osc-lib to build an
OSC-like stand-alone binary for plugins that would theoretically make
installing OSC optional for stand-alone client users.  This is not
complete yet, and as I said above, one reason I wish osc-lib had not
been merged into plugin requirements yet.  That said, as long as you
don't use those bits yet you will be fine, the utils, command, etc bits
are stable, it is the clientmanager and shell parts that are still being
developed.


Dean,

It seems like OSC now issues warnings if we use bits that are moved to 
osc-lib. Does it mean that now osc-lib is ready for all projects to 
switch to? If not, could you please revert the warnings? It's a bit 
confusing if we should our users warnings that we can't fix.


Thanks.



dt

--

Dean Troyer
dtro...@gmail.com 


__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev




__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Re: [openstack-dev] [mistral][osc-lib][openstackclient] is it too early for orc-lib?

[Dean Troyer]

On Mon, Jul 11, 2016 at 9:24 AM, Dmitry Tantsur  wrote:

> It seems like OSC now issues warnings if we use bits that are moved to
> osc-lib. Does it mean that now osc-lib is ready for all projects to switch
> to? If not, could you please revert the warnings? It's a bit confusing if
> we should our users warnings that we can't fix.
>

This only happens in OSC trunk, which will be released as 3.0.0 soon.  At
that point, osc-lib will also be released as 1.0 and be considered stable
and ready for general use.

dt

-- 

Dean Troyer
dtro...@gmail.com
__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev