You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@bigtop.apache.org by Roman Shaposhnik <ro...@shaposhnik.org> on 2017/03/17 15:16:27 UTC
Mergin ODPi changes into Bigtop
Hi!
as Evans quite rightfully noted right here:
https://issues.apache.org/jira/browse/BIGTOP-2696
I merged ODPi changes preserving ODPi
commit history.
Now, assuming that a given change coming from
ODPi looks OK to Bigtop development community,
what would be the preference here? Preserve the
history (this will mean, among other things, commit
messages with ODPi JIRA IDs in them, etc.) OR
squashing everything?
My rationale for preservation was my obsession with
IP lineage. Basically who contributed what at which
point under which License and ICLA. The downside
is the JIRA IDs and comments in the commits. Btw,
all those JIRAs are public, but of course they have nothing
to do with ASF.
Thoughts?
Thanks,
Roman.
Re: Mergin ODPi changes into Bigtop
Posted by Jay Vyas <ja...@gmail.com>.
I think the more info the better, and making extra commits that are noisy is ok. The priority is to keep big top and odp moving forward without making extra work for ODPi for contributing :).
> On Mar 17, 2017, at 11:16 AM, Roman Shaposhnik <ro...@shaposhnik.org> wrote:
>
> Hi!
>
> as Evans quite rightfully noted right here:
> https://issues.apache.org/jira/browse/BIGTOP-2696
> I merged ODPi changes preserving ODPi
> commit history.
>
> Now, assuming that a given change coming from
> ODPi looks OK to Bigtop development community,
> what would be the preference here? Preserve the
> history (this will mean, among other things, commit
> messages with ODPi JIRA IDs in them, etc.) OR
> squashing everything?
>
> My rationale for preservation was my obsession with
> IP lineage. Basically who contributed what at which
> point under which License and ICLA. The downside
> is the JIRA IDs and comments in the commits. Btw,
> all those JIRAs are public, but of course they have nothing
> to do with ASF.
>
> Thoughts?
>
> Thanks,
> Roman.
Re: Mergin ODPi changes into Bigtop
Posted by Roman Shaposhnik <ro...@shaposhnik.org>.
Hi Olaf!
not sure what you mean -- those 3 have one JIRA assigned to each of them.
See below:
On Mon, Mar 20, 2017 at 1:18 PM, Olaf Flebbe <of...@oflebbe.de> wrote:
> Cos,
>
> Surely you are referring to other commits, unless you are joking.
>
> https://github.com/apache/bigtop/commit/40fd25e52cace1245dd669307326c3cb20044335
ODPI-129
> https://github.com/apache/bigtop/commit/ba4c1bf7e29ba639137259308d8b6e0252ef7fb9
ODPI-100
> https://github.com/apache/bigtop/commit/d78693bed54fbb239aacff939a2e03663050d887
ODPI-72
Now, when we bring it into Bigtop -- there's no need to file 3 JIRAs
-- beacause the end
result is what we're after. In this sense it is no different from
somebody working on the
BIGTOP JIRA and having a few commits in their history. Most of the
time -- we're not interested
in how the person was fixing internal bugs -- we're interested in the
end result. That's why
we typically ask to squash. But there's not policy to force the squashing.
As I said -- I believe for ODPi's case we have a good reason not to --
but I'd be fine with
squashing.
Hence this thread ;-)
Thanks,
Roman.
Re: Mergin ODPi changes into Bigtop
Posted by Olaf Flebbe <of...@oflebbe.de>.
Cos,
Surely you are referring to other commits, unless you are joking.
https://github.com/apache/bigtop/commit/40fd25e52cace1245dd669307326c3cb20044335 <https://github.com/apache/bigtop/commit/40fd25e52cace1245dd669307326c3cb20044335>
https://github.com/apache/bigtop/commit/ba4c1bf7e29ba639137259308d8b6e0252ef7fb9 <https://github.com/apache/bigtop/commit/ba4c1bf7e29ba639137259308d8b6e0252ef7fb9>
https://github.com/apache/bigtop/commit/d78693bed54fbb239aacff939a2e03663050d887 <https://github.com/apache/bigtop/commit/d78693bed54fbb239aacff939a2e03663050d887>
Olaf
> Am 20.03.2017 um 06:00 schrieb Konstantin Boudnik <co...@boudnik.org>:
>
> On Sat, Mar 18, 2017 at 11:48 AM, Olaf Flebbe <of...@oflebbe.de> wrote:
>> Hi,
>>
>> Hooray! So we can smash the one commit == one JIRA mantra?
>
> That's not what I said, really. If you look into this ODPi
> contribution you'd notice that the same mantra applies there as well.
> When we ask github contributors to squash their commits, we actually
> going after noisy commits that reflect review comments and such. In
> other words, nothing that is material for the contribution.
>
> On the other hand, if you think this requirement should be more
> relaxed - let's discuss this as we usually do with this type of
> things.
>
> Regards,
> Cos
>
>> It was a bit of a pain in the $$$ to let the github commit's of contributors smashed to one commit. I am not a huge fan of this mantra, since it is a bit artifiically at times.
>>
>> Olaf
>>
>>
>>> Am 18.03.2017 um 19:10 schrieb Konstantin Boudnik <co...@apache.org>:
>>>
>>> Preserving the history is a good thing. This will clearly marks the merge
>>> point and not merely a "huge code blob coming from elsewhere". After all, one
>>> day it might be an interesting exercise for archaeologists, who knows ;)
>>>
>>> On Sat, Mar 18, 2017 at 02:23AM, Evans Ye wrote:
>>>> Roman you makes a good point about IP.
>>>>
>>>> I am thinking that is it possible to have a development guideline that one
>>>> ODPi commit can be one Bigtop commit? The thing here I am trying to propse
>>>
>>> I don't believe we ought to impose any sort of special guidelines for ODPi (or
>>> anyone else, in this matter). As we don't expect any other
>>> organization/individuals to do anything special outside of standard process
>>> for their contributions.
>>>
>>> We have project's development guidelines and anyone who feels like
>>> contributing shall follow these rules. For the merges like this we can roughly
>>> follow the principles of "code grants" and just accept it as is with the
>>> original history. But unlike code grants we don't need to do any special IP
>>> clearance or signing an official grant forms: ODPi development is happening
>>> under the same license, is happening in the open, and has similar governing
>>> principles as most of the Apache projects.
>>>
>>> Cos
>>>
>>>
>>>> is to bring two community working more closely from now on.
>>>> For those commits already made before, I think to preseve authors is better.
>>>>
>>>> Thanks for kicking off the discussion.
>>>>
>>>>
>>>> 김영우 (YoungWoo Kim) <yw...@apache.org>於 2017年3月18日 週六，上午9:06寫道：
>>>>
>>>>> Personal preference is merging into one commit. But in this case, I believe
>>>>>
>>>>> it's good to go with preserving the commit history. The history info
>>>>>
>>>>> provides us the details of development and also commits are already
>>>>>
>>>>> reviewed and tested by bigtop members right? :-)
>>>>>
>>>>>
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Youngwoo
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> 2017. 3. 18. 오전 12:16에 "Roman Shaposhnik" <ro...@shaposhnik.org>님이 작성:
>>>>>
>>>>>
>>>>>
>>>>>> Hi!
>>>>>
>>>>>>
>>>>>
>>>>>> as Evans quite rightfully noted right here:
>>>>>
>>>>>> https://issues.apache.org/jira/browse/BIGTOP-2696
>>>>>
>>>>>> I merged ODPi changes preserving ODPi
>>>>>
>>>>>> commit history.
>>>>>
>>>>>>
>>>>>
>>>>>> Now, assuming that a given change coming from
>>>>>
>>>>>> ODPi looks OK to Bigtop development community,
>>>>>
>>>>>> what would be the preference here? Preserve the
>>>>>
>>>>>> history (this will mean, among other things, commit
>>>>>
>>>>>> messages with ODPi JIRA IDs in them, etc.) OR
>>>>>
>>>>>> squashing everything?
>>>>>
>>>>>>
>>>>>
>>>>>> My rationale for preservation was my obsession with
>>>>>
>>>>>> IP lineage. Basically who contributed what at which
>>>>>
>>>>>> point under which License and ICLA. The downside
>>>>>
>>>>>> is the JIRA IDs and comments in the commits. Btw,
>>>>>
>>>>>> all those JIRAs are public, but of course they have nothing
>>>>>
>>>>>> to do with ASF.
>>>>>
>>>>>>
>>>>>
>>>>>> Thoughts?
>>>>>
>>>>>>
>>>>>
>>>>>> Thanks,
>>>>>
>>>>>> Roman.
>>>>>
>>>>>>
>>>>>
>>>>>
>>
Re: Mergin ODPi changes into Bigtop
Posted by Konstantin Boudnik <co...@boudnik.org>.
On Sat, Mar 18, 2017 at 11:48 AM, Olaf Flebbe <of...@oflebbe.de> wrote:
> Hi,
>
> Hooray! So we can smash the one commit == one JIRA mantra?
That's not what I said, really. If you look into this ODPi
contribution you'd notice that the same mantra applies there as well.
When we ask github contributors to squash their commits, we actually
going after noisy commits that reflect review comments and such. In
other words, nothing that is material for the contribution.
On the other hand, if you think this requirement should be more
relaxed - let's discuss this as we usually do with this type of
things.
Regards,
Cos
> It was a bit of a pain in the $$$ to let the github commit's of contributors smashed to one commit. I am not a huge fan of this mantra, since it is a bit artifiically at times.
>
> Olaf
>
>
>> Am 18.03.2017 um 19:10 schrieb Konstantin Boudnik <co...@apache.org>:
>>
>> Preserving the history is a good thing. This will clearly marks the merge
>> point and not merely a "huge code blob coming from elsewhere". After all, one
>> day it might be an interesting exercise for archaeologists, who knows ;)
>>
>> On Sat, Mar 18, 2017 at 02:23AM, Evans Ye wrote:
>>> Roman you makes a good point about IP.
>>>
>>> I am thinking that is it possible to have a development guideline that one
>>> ODPi commit can be one Bigtop commit? The thing here I am trying to propse
>>
>> I don't believe we ought to impose any sort of special guidelines for ODPi (or
>> anyone else, in this matter). As we don't expect any other
>> organization/individuals to do anything special outside of standard process
>> for their contributions.
>>
>> We have project's development guidelines and anyone who feels like
>> contributing shall follow these rules. For the merges like this we can roughly
>> follow the principles of "code grants" and just accept it as is with the
>> original history. But unlike code grants we don't need to do any special IP
>> clearance or signing an official grant forms: ODPi development is happening
>> under the same license, is happening in the open, and has similar governing
>> principles as most of the Apache projects.
>>
>> Cos
>>
>>
>>> is to bring two community working more closely from now on.
>>> For those commits already made before, I think to preseve authors is better.
>>>
>>> Thanks for kicking off the discussion.
>>>
>>>
>>> 김영우 (YoungWoo Kim) <yw...@apache.org>於 2017年3月18日 週六，上午9:06寫道：
>>>
>>>> Personal preference is merging into one commit. But in this case, I believe
>>>>
>>>> it's good to go with preserving the commit history. The history info
>>>>
>>>> provides us the details of development and also commits are already
>>>>
>>>> reviewed and tested by bigtop members right? :-)
>>>>
>>>>
>>>>
>>>> Thanks,
>>>>
>>>> Youngwoo
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> 2017. 3. 18. 오전 12:16에 "Roman Shaposhnik" <ro...@shaposhnik.org>님이 작성:
>>>>
>>>>
>>>>
>>>>> Hi!
>>>>
>>>>>
>>>>
>>>>> as Evans quite rightfully noted right here:
>>>>
>>>>> https://issues.apache.org/jira/browse/BIGTOP-2696
>>>>
>>>>> I merged ODPi changes preserving ODPi
>>>>
>>>>> commit history.
>>>>
>>>>>
>>>>
>>>>> Now, assuming that a given change coming from
>>>>
>>>>> ODPi looks OK to Bigtop development community,
>>>>
>>>>> what would be the preference here? Preserve the
>>>>
>>>>> history (this will mean, among other things, commit
>>>>
>>>>> messages with ODPi JIRA IDs in them, etc.) OR
>>>>
>>>>> squashing everything?
>>>>
>>>>>
>>>>
>>>>> My rationale for preservation was my obsession with
>>>>
>>>>> IP lineage. Basically who contributed what at which
>>>>
>>>>> point under which License and ICLA. The downside
>>>>
>>>>> is the JIRA IDs and comments in the commits. Btw,
>>>>
>>>>> all those JIRAs are public, but of course they have nothing
>>>>
>>>>> to do with ASF.
>>>>
>>>>>
>>>>
>>>>> Thoughts?
>>>>
>>>>>
>>>>
>>>>> Thanks,
>>>>
>>>>> Roman.
>>>>
>>>>>
>>>>
>>>>
>
Re: Mergin ODPi changes into Bigtop
Posted by Olaf Flebbe <of...@oflebbe.de>.
Hi,
Hooray! So we can smash the one commit == one JIRA mantra?
It was a bit of a pain in the $$$ to let the github commit's of contributors smashed to one commit. I am not a huge fan of this mantra, since it is a bit artifiically at times.
Olaf
> Am 18.03.2017 um 19:10 schrieb Konstantin Boudnik <co...@apache.org>:
>
> Preserving the history is a good thing. This will clearly marks the merge
> point and not merely a "huge code blob coming from elsewhere". After all, one
> day it might be an interesting exercise for archaeologists, who knows ;)
>
> On Sat, Mar 18, 2017 at 02:23AM, Evans Ye wrote:
>> Roman you makes a good point about IP.
>>
>> I am thinking that is it possible to have a development guideline that one
>> ODPi commit can be one Bigtop commit? The thing here I am trying to propse
>
> I don't believe we ought to impose any sort of special guidelines for ODPi (or
> anyone else, in this matter). As we don't expect any other
> organization/individuals to do anything special outside of standard process
> for their contributions.
>
> We have project's development guidelines and anyone who feels like
> contributing shall follow these rules. For the merges like this we can roughly
> follow the principles of "code grants" and just accept it as is with the
> original history. But unlike code grants we don't need to do any special IP
> clearance or signing an official grant forms: ODPi development is happening
> under the same license, is happening in the open, and has similar governing
> principles as most of the Apache projects.
>
> Cos
>
>
>> is to bring two community working more closely from now on.
>> For those commits already made before, I think to preseve authors is better.
>>
>> Thanks for kicking off the discussion.
>>
>>
>> 김영우 (YoungWoo Kim) <yw...@apache.org>於 2017年3月18日 週六，上午9:06寫道：
>>
>>> Personal preference is merging into one commit. But in this case, I believe
>>>
>>> it's good to go with preserving the commit history. The history info
>>>
>>> provides us the details of development and also commits are already
>>>
>>> reviewed and tested by bigtop members right? :-)
>>>
>>>
>>>
>>> Thanks,
>>>
>>> Youngwoo
>>>
>>>
>>>
>>>
>>>
>>> 2017. 3. 18. 오전 12:16에 "Roman Shaposhnik" <ro...@shaposhnik.org>님이 작성:
>>>
>>>
>>>
>>>> Hi!
>>>
>>>>
>>>
>>>> as Evans quite rightfully noted right here:
>>>
>>>> https://issues.apache.org/jira/browse/BIGTOP-2696
>>>
>>>> I merged ODPi changes preserving ODPi
>>>
>>>> commit history.
>>>
>>>>
>>>
>>>> Now, assuming that a given change coming from
>>>
>>>> ODPi looks OK to Bigtop development community,
>>>
>>>> what would be the preference here? Preserve the
>>>
>>>> history (this will mean, among other things, commit
>>>
>>>> messages with ODPi JIRA IDs in them, etc.) OR
>>>
>>>> squashing everything?
>>>
>>>>
>>>
>>>> My rationale for preservation was my obsession with
>>>
>>>> IP lineage. Basically who contributed what at which
>>>
>>>> point under which License and ICLA. The downside
>>>
>>>> is the JIRA IDs and comments in the commits. Btw,
>>>
>>>> all those JIRAs are public, but of course they have nothing
>>>
>>>> to do with ASF.
>>>
>>>>
>>>
>>>> Thoughts?
>>>
>>>>
>>>
>>>> Thanks,
>>>
>>>> Roman.
>>>
>>>>
>>>
>>>
Re: Mergin ODPi changes into Bigtop
Posted by Konstantin Boudnik <co...@apache.org>.
Preserving the history is a good thing. This will clearly marks the merge
point and not merely a "huge code blob coming from elsewhere". After all, one
day it might be an interesting exercise for archaeologists, who knows ;)
On Sat, Mar 18, 2017 at 02:23AM, Evans Ye wrote:
> Roman you makes a good point about IP.
>
> I am thinking that is it possible to have a development guideline that one
> ODPi commit can be one Bigtop commit? The thing here I am trying to propse
I don't believe we ought to impose any sort of special guidelines for ODPi (or
anyone else, in this matter). As we don't expect any other
organization/individuals to do anything special outside of standard process
for their contributions.
We have project's development guidelines and anyone who feels like
contributing shall follow these rules. For the merges like this we can roughly
follow the principles of "code grants" and just accept it as is with the
original history. But unlike code grants we don't need to do any special IP
clearance or signing an official grant forms: ODPi development is happening
under the same license, is happening in the open, and has similar governing
principles as most of the Apache projects.
Cos
> is to bring two community working more closely from now on.
> For those commits already made before, I think to preseve authors is better.
>
> Thanks for kicking off the discussion.
>
>
> 김영우 (YoungWoo Kim) <yw...@apache.org>於 2017年3月18日 週六，上午9:06寫道：
>
> > Personal preference is merging into one commit. But in this case, I believe
> >
> > it's good to go with preserving the commit history. The history info
> >
> > provides us the details of development and also commits are already
> >
> > reviewed and tested by bigtop members right? :-)
> >
> >
> >
> > Thanks,
> >
> > Youngwoo
> >
> >
> >
> >
> >
> > 2017. 3. 18. 오전 12:16에 "Roman Shaposhnik" <ro...@shaposhnik.org>님이 작성:
> >
> >
> >
> > > Hi!
> >
> > >
> >
> > > as Evans quite rightfully noted right here:
> >
> > > https://issues.apache.org/jira/browse/BIGTOP-2696
> >
> > > I merged ODPi changes preserving ODPi
> >
> > > commit history.
> >
> > >
> >
> > > Now, assuming that a given change coming from
> >
> > > ODPi looks OK to Bigtop development community,
> >
> > > what would be the preference here? Preserve the
> >
> > > history (this will mean, among other things, commit
> >
> > > messages with ODPi JIRA IDs in them, etc.) OR
> >
> > > squashing everything?
> >
> > >
> >
> > > My rationale for preservation was my obsession with
> >
> > > IP lineage. Basically who contributed what at which
> >
> > > point under which License and ICLA. The downside
> >
> > > is the JIRA IDs and comments in the commits. Btw,
> >
> > > all those JIRAs are public, but of course they have nothing
> >
> > > to do with ASF.
> >
> > >
> >
> > > Thoughts?
> >
> > >
> >
> > > Thanks,
> >
> > > Roman.
> >
> > >
> >
> >
Re: Mergin ODPi changes into Bigtop
Posted by Evans Ye <ev...@apache.org>.
Roman you makes a good point about IP.
I am thinking that is it possible to have a development guideline that one
ODPi commit can be one Bigtop commit? The thing here I am trying to propse
is to bring two community working more closely from now on.
For those commits already made before, I think to preseve authors is better.
Thanks for kicking off the discussion.
김영우 (YoungWoo Kim) <yw...@apache.org>於 2017年3月18日 週六，上午9:06寫道：
> Personal preference is merging into one commit. But in this case, I believe
>
> it's good to go with preserving the commit history. The history info
>
> provides us the details of development and also commits are already
>
> reviewed and tested by bigtop members right? :-)
>
>
>
> Thanks,
>
> Youngwoo
>
>
>
>
>
> 2017. 3. 18. 오전 12:16에 "Roman Shaposhnik" <ro...@shaposhnik.org>님이 작성:
>
>
>
> > Hi!
>
> >
>
> > as Evans quite rightfully noted right here:
>
> > https://issues.apache.org/jira/browse/BIGTOP-2696
>
> > I merged ODPi changes preserving ODPi
>
> > commit history.
>
> >
>
> > Now, assuming that a given change coming from
>
> > ODPi looks OK to Bigtop development community,
>
> > what would be the preference here? Preserve the
>
> > history (this will mean, among other things, commit
>
> > messages with ODPi JIRA IDs in them, etc.) OR
>
> > squashing everything?
>
> >
>
> > My rationale for preservation was my obsession with
>
> > IP lineage. Basically who contributed what at which
>
> > point under which License and ICLA. The downside
>
> > is the JIRA IDs and comments in the commits. Btw,
>
> > all those JIRAs are public, but of course they have nothing
>
> > to do with ASF.
>
> >
>
> > Thoughts?
>
> >
>
> > Thanks,
>
> > Roman.
>
> >
>
>
Re: Mergin ODPi changes into Bigtop
Posted by "김영우 (YoungWoo Kim)" <yw...@apache.org>.
Personal preference is merging into one commit. But in this case, I believe
it's good to go with preserving the commit history. The history info
provides us the details of development and also commits are already
reviewed and tested by bigtop members right? :-)
Thanks,
Youngwoo
2017. 3. 18. 오전 12:16에 "Roman Shaposhnik" <ro...@shaposhnik.org>님이 작성:
> Hi!
>
> as Evans quite rightfully noted right here:
> https://issues.apache.org/jira/browse/BIGTOP-2696
> I merged ODPi changes preserving ODPi
> commit history.
>
> Now, assuming that a given change coming from
> ODPi looks OK to Bigtop development community,
> what would be the preference here? Preserve the
> history (this will mean, among other things, commit
> messages with ODPi JIRA IDs in them, etc.) OR
> squashing everything?
>
> My rationale for preservation was my obsession with
> IP lineage. Basically who contributed what at which
> point under which License and ICLA. The downside
> is the JIRA IDs and comments in the commits. Btw,
> all those JIRAs are public, but of course they have nothing
> to do with ASF.
>
> Thoughts?
>
> Thanks,
> Roman.
>