Re: How to ease contributions to chemisty-cmsilib

[Mignon, Laurent]

On Mon, Oct 30, 2017 at 2:40 PM, Jeff Potts  wrote:

> >
> > Unfortunately, I do not have sufficient rights on
> > the repository to access the configuration parameters. Does anyone have
> the
> > rights to activate this integration?
>
>
> Is this referring to the GitHub repo? If so, I think that has to go through
> the Apache Infrastructure group but I do not know for sure.
>
>  Yes I'm referring to the Github repo.

Laurent

Re: How to ease contributions to chemisty-cmsilib

[Jeff Potts]

>
> Unfortunately, I do not have sufficient rights on
> the repository to access the configuration parameters. Does anyone have the
> rights to activate this integration?


Is this referring to the GitHub repo? If so, I think that has to go through
the Apache Infrastructure group but I do not know for sure.

Jeff

On Mon, Oct 30, 2017 at 7:44 AM, Mignon, Laurent 
wrote:

> Hi,
>
> I've refactored the testsuite of cmislib and put in place a CI based on
> travis.
> All the changes are available in a PR on github.
> https://github.com/apache/chemistry-cmislib/pull/10
> Contrary to what was said previously there is no message sent to the ML
> when a PR is created on github.
> It also seems that the integration with travis is not enabled by default on
> the project in github. Unfortunately, I do not have sufficient rights on
> the repository to access the configuration parameters. Does anyone have the
> rights to activate this integration? In the meantime, you can see the
> results of the CI for this PR tests here: https: //
> travis-ci.org/acsone/chemistry-cmislib/builds/292630173
>
> Regards,
>
> Laurent Mignon
>
>
> On Fri, Oct 20, 2017 at 3:44 PM, Jeff Potts  wrote:
>
> > We actually discussed this quite a while ago and decided it was okay to
> > move cmislib to git. However, upon looking at it, I realized that the
> only
> > difference was going to be the command line tool developers used. It was
> > not going to make the project available on github (beyond what is already
> > available with the mirror) and it didn't look like it was going to ease
> > contributions through pull requests.
> >
> > So I am supportive of moving to git, even if it only means that the
> people
> > who contribute can use git instead of svn.
> >
> > But, I am curious as to exactly how the move will make the project more
> > visible, make changes more visible, and make the review process simpler.
> >
> > Jeff
> >
> > On Fri, Oct 20, 2017 at 4:51 AM, Mignon, Laurent <
> laurent.mig...@acsone.eu
> > >
> > wrote:
> >
> > > On Fri, Oct 20, 2017 at 11:07 AM, Nick Burch 
> > wrote:
> > >
> > > > On Fri, 20 Oct 2017, Florian Müller wrote:
> > > >
> > > >> I like the idea of moving all Apache Chemistry projects to Git. It's
> > > some
> > > >> work, though. Someone has to drive it...
> > > >>
> > > >
> > > > I think you already have github mirrors for all of them!
> > > > https://github.com/apache?utf8=%E2%9C%93=chemistry==
> > > >
> > > >
> > > I know that we already have a mirror but it's a mirror and you always
> > have
> > > to make the work twice.
> > >
> > > (If there's any missing just ask infra)
> > > >
> > > >
> > > > We switched Apache Tika to git a year or so ago, and it's a
> non-trivial
> > > > amount of work, so you'd want to ensure you're actually getting
> benefit
> > > > from doing it...
> > > >
> > > > Nick
> > >
> > >
> > > What are the impacts? From what I find as information on
> > chemistry-cmislib
> > > it's difficult to assess the workload that this represents. The only
> > > information I can find is that you have to make a ticket to the
> support.
> > >
> > > The ASF provides CI servers which run on svn or git commits, you just
> > have
> > > > to ask for one to be turned on. The main two offerings are:
> > > >  * https://cwiki.apache.org/confluence/display/INFRA/Jenkins
> > > >  * https://ci.apache.org/buildbot.html
> > >
> > >
> > > To automate the tests of cmislib we must be able to install/access a
> cmis
> > > server (alfresco, nuxeo, ???). Do we have this kind of server available
> > > somewhere? With travis or gitlabci, I know  it's possible to install
> this
> > > kind of server before running the tests or to use a container with all
> > the
> > > required stuff pre installed. Is-it possible with the ASF CI servers?
> > >
> > > Another advantage with travis is that configuring your build is part of
> > the
> > > project sources. It is therefore possible to adapt it from one branch
> to
> > > another. This is a feature that is very important to me in anticipation
> > of
> > > adding python3 support. I know that for buildbot, this is not possible.
> > > It's too long since I've used travis to know if it's possible with this
> > > one.
> > >
> > > lmignon
> > >
> >
>

Re: How to ease contributions to chemisty-cmsilib

[Mignon, Laurent]

Hi,

I've refactored the testsuite of cmislib and put in place a CI based on
travis.
All the changes are available in a PR on github.
https://github.com/apache/chemistry-cmislib/pull/10
Contrary to what was said previously there is no message sent to the ML
when a PR is created on github.
It also seems that the integration with travis is not enabled by default on
the project in github. Unfortunately, I do not have sufficient rights on
the repository to access the configuration parameters. Does anyone have the
rights to activate this integration? In the meantime, you can see the
results of the CI for this PR tests here: https: //
travis-ci.org/acsone/chemistry-cmislib/builds/292630173

Regards,

Laurent Mignon


On Fri, Oct 20, 2017 at 3:44 PM, Jeff Potts  wrote:

> We actually discussed this quite a while ago and decided it was okay to
> move cmislib to git. However, upon looking at it, I realized that the only
> difference was going to be the command line tool developers used. It was
> not going to make the project available on github (beyond what is already
> available with the mirror) and it didn't look like it was going to ease
> contributions through pull requests.
>
> So I am supportive of moving to git, even if it only means that the people
> who contribute can use git instead of svn.
>
> But, I am curious as to exactly how the move will make the project more
> visible, make changes more visible, and make the review process simpler.
>
> Jeff
>
> On Fri, Oct 20, 2017 at 4:51 AM, Mignon, Laurent  >
> wrote:
>
> > On Fri, Oct 20, 2017 at 11:07 AM, Nick Burch 
> wrote:
> >
> > > On Fri, 20 Oct 2017, Florian Müller wrote:
> > >
> > >> I like the idea of moving all Apache Chemistry projects to Git. It's
> > some
> > >> work, though. Someone has to drive it...
> > >>
> > >
> > > I think you already have github mirrors for all of them!
> > > https://github.com/apache?utf8=%E2%9C%93=chemistry==
> > >
> > >
> > I know that we already have a mirror but it's a mirror and you always
> have
> > to make the work twice.
> >
> > (If there's any missing just ask infra)
> > >
> > >
> > > We switched Apache Tika to git a year or so ago, and it's a non-trivial
> > > amount of work, so you'd want to ensure you're actually getting benefit
> > > from doing it...
> > >
> > > Nick
> >
> >
> > What are the impacts? From what I find as information on
> chemistry-cmislib
> > it's difficult to assess the workload that this represents. The only
> > information I can find is that you have to make a ticket to the support.
> >
> > The ASF provides CI servers which run on svn or git commits, you just
> have
> > > to ask for one to be turned on. The main two offerings are:
> > >  * https://cwiki.apache.org/confluence/display/INFRA/Jenkins
> > >  * https://ci.apache.org/buildbot.html
> >
> >
> > To automate the tests of cmislib we must be able to install/access a cmis
> > server (alfresco, nuxeo, ???). Do we have this kind of server available
> > somewhere? With travis or gitlabci, I know  it's possible to install this
> > kind of server before running the tests or to use a container with all
> the
> > required stuff pre installed. Is-it possible with the ASF CI servers?
> >
> > Another advantage with travis is that configuring your build is part of
> the
> > project sources. It is therefore possible to adapt it from one branch to
> > another. This is a feature that is very important to me in anticipation
> of
> > adding python3 support. I know that for buildbot, this is not possible.
> > It's too long since I've used travis to know if it's possible with this
> > one.
> >
> > lmignon
> >
>

Re: How to ease contributions to chemisty-cmsilib

[Jeff Potts]

We actually discussed this quite a while ago and decided it was okay to
move cmislib to git. However, upon looking at it, I realized that the only
difference was going to be the command line tool developers used. It was
not going to make the project available on github (beyond what is already
available with the mirror) and it didn't look like it was going to ease
contributions through pull requests.

So I am supportive of moving to git, even if it only means that the people
who contribute can use git instead of svn.

But, I am curious as to exactly how the move will make the project more
visible, make changes more visible, and make the review process simpler.

Jeff

On Fri, Oct 20, 2017 at 4:51 AM, Mignon, Laurent 
wrote:

> On Fri, Oct 20, 2017 at 11:07 AM, Nick Burch  wrote:
>
> > On Fri, 20 Oct 2017, Florian Müller wrote:
> >
> >> I like the idea of moving all Apache Chemistry projects to Git. It's
> some
> >> work, though. Someone has to drive it...
> >>
> >
> > I think you already have github mirrors for all of them!
> > https://github.com/apache?utf8=%E2%9C%93=chemistry==
> >
> >
> I know that we already have a mirror but it's a mirror and you always have
> to make the work twice.
>
> (If there's any missing just ask infra)
> >
> >
> > We switched Apache Tika to git a year or so ago, and it's a non-trivial
> > amount of work, so you'd want to ensure you're actually getting benefit
> > from doing it...
> >
> > Nick
>
>
> What are the impacts? From what I find as information on chemistry-cmislib
> it's difficult to assess the workload that this represents. The only
> information I can find is that you have to make a ticket to the support.
>
> The ASF provides CI servers which run on svn or git commits, you just have
> > to ask for one to be turned on. The main two offerings are:
> >  * https://cwiki.apache.org/confluence/display/INFRA/Jenkins
> >  * https://ci.apache.org/buildbot.html
>
>
> To automate the tests of cmislib we must be able to install/access a cmis
> server (alfresco, nuxeo, ???). Do we have this kind of server available
> somewhere? With travis or gitlabci, I know  it's possible to install this
> kind of server before running the tests or to use a container with all the
> required stuff pre installed. Is-it possible with the ASF CI servers?
>
> Another advantage with travis is that configuring your build is part of the
> project sources. It is therefore possible to adapt it from one branch to
> another. This is a feature that is very important to me in anticipation of
> adding python3 support. I know that for buildbot, this is not possible.
> It's too long since I've used travis to know if it's possible with this
> one.
>
> lmignon
>

Re: How to ease contributions to chemisty-cmsilib

[Mignon, Laurent]

On Fri, Oct 20, 2017 at 11:07 AM, Nick Burch  wrote:

> On Fri, 20 Oct 2017, Florian Müller wrote:
>
>> I like the idea of moving all Apache Chemistry projects to Git. It's some
>> work, though. Someone has to drive it...
>>
>
> I think you already have github mirrors for all of them!
> https://github.com/apache?utf8=%E2%9C%93=chemistry==
>
>
I know that we already have a mirror but it's a mirror and you always have
to make the work twice.

(If there's any missing just ask infra)
>
>
> We switched Apache Tika to git a year or so ago, and it's a non-trivial
> amount of work, so you'd want to ensure you're actually getting benefit
> from doing it...
>
> Nick


What are the impacts? From what I find as information on chemistry-cmislib
it's difficult to assess the workload that this represents. The only
information I can find is that you have to make a ticket to the support.

The ASF provides CI servers which run on svn or git commits, you just have
> to ask for one to be turned on. The main two offerings are:
>  * https://cwiki.apache.org/confluence/display/INFRA/Jenkins
>  * https://ci.apache.org/buildbot.html


To automate the tests of cmislib we must be able to install/access a cmis
server (alfresco, nuxeo, ???). Do we have this kind of server available
somewhere? With travis or gitlabci, I know  it's possible to install this
kind of server before running the tests or to use a container with all the
required stuff pre installed. Is-it possible with the ASF CI servers?

Another advantage with travis is that configuring your build is part of the
project sources. It is therefore possible to adapt it from one branch to
another. This is a feature that is very important to me in anticipation of
adding python3 support. I know that for buildbot, this is not possible.
It's too long since I've used travis to know if it's possible with this one.

lmignon

Re: How to ease contributions to chemisty-cmsilib

[Nick Burch]

On Fri, 20 Oct 2017, Florian Müller wrote:
I like the idea of moving all Apache Chemistry projects to Git. It's 
some work, though. Someone has to drive it...


I think you already have github mirrors for all of them!
https://github.com/apache?utf8=%E2%9C%93=chemistry==

(If there's any missing just ask infra)


We switched Apache Tika to git a year or so ago, and it's a non-trivial 
amount of work, so you'd want to ensure you're actually getting benefit 
from doing it...


Nick

Re: How to ease contributions to chemisty-cmsilib

[Nick Burch]

On Fri, 20 Oct 2017, Laurent Mignon wrote:

In discussions with Jeff and Florian on this topic, Florian mentioned that
the ASF now offers a tool for integrating projects with Github. (
https://gitbox.apache.org/)


There's already a read-only mirror of cmislib (and the other Chemistry 
projects) on github - https://github.com/apache/chemistry-cmislib


Anyone can clone from that, and open pull requests against it. Pull 
requests get emailed to the dev list, and if you apply the patch in svn 
with the right commit message they'll get auto-closed



I can't see any mention of that on the chemistry website, and the source 
code link just dumps you straight into svn


As a first step, I'd suggest you change the "source code" link to be a 
page which explains about the github mirror, suggests people use it, as 
well as providing the svn url. Something like 
http://poi.apache.org/subversion.html might work




This change would allow:
* A better visibility of the project
* A better publicity of ongoing changes and improvements
* A simpler review process open to everyone


You should have all of those already via the current github mirror!


* The use of continuous integration servers such as' TravisCI' and tools
for analysis and visualization of code quality such as' CodeClimate'.


The ASF provides CI servers which run on svn or git commits, you just have 
to ask for one to be turned on. The main two offerings are:

 * https://cwiki.apache.org/confluence/display/INFRA/Jenkins
 * https://ci.apache.org/buildbot.html


That's not to say that you shouldn't move to "git dual hosting", just that 
I'd suggest you make sure you'd be moving for the right reasons!


Nick

Re: How to ease contributions to chemisty-cmsilib

[Florian Müller]

I like the idea of moving all Apache Chemistry projects to Git.
It's some work, though. Someone has to drive it...

- Florian


+1 from me, although I don't contribute to cmislib, and as a matter of 
fact

I would also support moving the Java chemistry codebase to git.

Florent

On Fri, Oct 20, 2017 at 9:31 AM, Laurent Mignon  
wrote:



Hi,

As a new committer I confess that it is not easy to find the process 
to
follow to propose new improvements or bug fix to chemistry-cmislib and 
to

allow the code review by others.

In recent years, tools such as gitlab, github, bitbucket,... have 
emerged

and all propose a simple way to organize, manage and support project
developments.

In discussions with Jeff and Florian on this topic, Florian mentioned 
that

the ASF now offers a tool for integrating projects with Github. (
https://gitbox.apache.org/)

As a regular user of these tools for others OS projects I would like 
to

propose moving the cmislib project to GIT/github.

This change would allow:
* A better visibility of the project
* A better publicity of ongoing changes and improvements
* A simpler review process open to everyone
* The use of continuous integration servers such as' TravisCI' and 
tools

for analysis and visualization of code quality such as' CodeClimate'.

I also think that the use of tools widely used by the python community
could make the project more attractive.

What do you think about that?

Regards,

Laurent

Re: How to ease contributions to chemisty-cmsilib

[Florent Guillaume]

+1 from me, although I don't contribute to cmislib, and as a matter of fact
I would also support moving the Java chemistry codebase to git.

Florent

On Fri, Oct 20, 2017 at 9:31 AM, Laurent Mignon  wrote:

> Hi,
>
> As a new committer I confess that it is not easy to find the process to
> follow to propose new improvements or bug fix to chemistry-cmislib and to
> allow the code review by others.
>
> In recent years, tools such as gitlab, github, bitbucket,... have emerged
> and all propose a simple way to organize, manage and support project
> developments.
>
> In discussions with Jeff and Florian on this topic, Florian mentioned that
> the ASF now offers a tool for integrating projects with Github. (
> https://gitbox.apache.org/)
>
> As a regular user of these tools for others OS projects I would like to
> propose moving the cmislib project to GIT/github.
>
> This change would allow:
> * A better visibility of the project
> * A better publicity of ongoing changes and improvements
> * A simpler review process open to everyone
> * The use of continuous integration servers such as' TravisCI' and tools
> for analysis and visualization of code quality such as' CodeClimate'.
>
> I also think that the use of tools widely used by the python community
> could make the project more attractive.
>
> What do you think about that?
>
> Regards,
>
> Laurent
>



-- 
[image: Nuxeo]

Florent Guillaume
Head of R

Twitter: @efge