Re: [Bioc-devel] Git Transition Plan

[Turaga, Nitesh]

Hi Philipp,

The beta testing phase just started. I sent out an email to the bio-devel 
mailing list today.

We thank you for your enthusiasm to participate in this beta test phase.

Best Regards,

Nitesh


hi, do you know by now when the beta will start? i�m honestly a bit tired of 
SVN and would like to participate


Von: "Turaga, Nitesh" 
<nitesh.tur...@roswellpark.org<mailto:nitesh.tur...@roswellpark.org>>
An: bioc-devel@r-project.org<mailto:bioc-devel@r-project.org>
Gesendet: Donnerstag, 1. Juni 2017 14:19:59
Betreff: Re: [Bioc-devel] Git Transition Plan

Hi Nathan,

We�ve been working on a few things for the beta phase of the transition.

1. Getting the experiment data packages on board
2. Making sure manifests are in order along with their commit history.
3. Writing documentation for our new workflows for developers/maintainers.

The beta process has not started but, we expect it to be in play very
soon.

New package submissions are still going through SVN. We will make sure to
send out an email/ post in the support group NEWS, and inform all the
users as soon as that happens.

You did not miss anything.

Best,

Nitesh




>Hi Nitesh and all,
>
>Can you give any status update on the SVN-to-git transition? I haven't
>seen any news since this exchange in March -- is the announced public
>beta in progress or ongoing? Are new package submissions going straight
>into git at this point?
>
>Or did I miss something?
>
>Thanks,
>
>Nathan
>
>
>On 04/01/2017 09:14 AM, Martin Morgan wrote:
>> On 03/29/2017 03:55 PM, Henrik Bengtsson wrote:
>>> Thanks.  I have a few thoughts and questions in order to plan ahead:
>>>
>>>> - Our plan is to make a 'clean' transition from SVN to git,
>>>> approximately one month after the next Bioconductor release.
>>>> Developers or users will not have access to the SVN system after the
>>>> date of transition.
>>>
>>> In order to preserve commit authorship, what's your plan for mapping
>>> SVN username to Git 'user.email' and 'user.name'?  The 'user.email' is
>>> what GitHub uses to associate commits and contributions to GitHub
>>> accounts.
>>
>> The svn administrators kept a comprehensive (but not complete) record
>> of svn id, real user name, and contact email at the time the svn id
>> was create. We use this to map between svn commits and git user name
>> and email address.
>>
>> The information is not entirely consistent, with some fields for some
>> records 'missing' (e.g., my record doesn't contain my real name) and
>> of course out of date (e.g., my email address).
>>
>> Our intention is NOT to re-write history, but to map the information
>> that we have to the git repositories. So my svn commits appear without
>> a proper name, and with an outdated email address. Of course, new
>> commits after the transition will contain whatever info git provides.
>>
>>>
>>> BTW, using obsolete email addresses may prevent people from being
>>> associated with those email addresses on GitHub and other online
>>> services that require authentication of authorship claims (which go
>>> out via those email addresses).
>>>
>>> When I SVN-to-Git exported my Bioconductor packages a few years ago, I
>>> could handle manually because there were not too many contributors in
>>> the SVN logs and I reached out to each of them and asked what email
>>> addresses they would prefer to have in the Git commits.  That approach
>>> is obviously not feasible to automate for all Bioconductor packages.
>>> Maybe this can be handled optionally by each package maintainers by
>>> adding a .gitauthors file to the package root (or possibly via a
>>> global Bioc one that everyone can commit to), e.g.
>>> ```r
>>> hb = Henrik Bengtsson <henr...@braju.com<mailto:henr...@braju.com>>
>>> j...@foo.com<mailto:j...@foo.com> = John Doe 
>>> <j...@someone.org<mailto:j...@someone.org>>
>>> ```
>>> and then Bioc can use this mapping when exporting to Git?
>>>
>>> Finally, will people like me who already done the SVN-to-Git migration
>>> be able to use that instead of the Bioc generated one? (I assume not,
>>> but worth asking)
>>
>> No, we'll make a git snapshots of svn at the time of transition, and
>> these git repositories will be the cannonical version -- developers
>> will need to sync with these as they see fit.
>>
>> Martin
>>
>>>
>>> Thanks,
>>>
>>> Henrik
>>>
>>> On W

Re: [Bioc-devel] Git Transition Plan

[Angerer, Philipp]

hi, do you know by now when the beta will start? i’m honestly a bit tired of 
SVN and would like to participate 


Von: "Turaga, Nitesh" <nitesh.tur...@roswellpark.org> 
An: bioc-devel@r-project.org 
Gesendet: Donnerstag, 1. Juni 2017 14:19:59 
Betreff: Re: [Bioc-devel] Git Transition Plan 

Hi Nathan, 

We¹ve been working on a few things for the beta phase of the transition. 

1. Getting the experiment data packages on board 
2. Making sure manifests are in order along with their commit history. 
3. Writing documentation for our new workflows for developers/maintainers. 

The beta process has not started but, we expect it to be in play very 
soon. 

New package submissions are still going through SVN. We will make sure to 
send out an email/ post in the support group NEWS, and inform all the 
users as soon as that happens. 

You did not miss anything. 

Best, 

Nitesh 




>Hi Nitesh and all, 
> 
>Can you give any status update on the SVN-to-git transition? I haven't 
>seen any news since this exchange in March -- is the announced public 
>beta in progress or ongoing? Are new package submissions going straight 
>into git at this point? 
> 
>Or did I miss something? 
> 
>Thanks, 
> 
>Nathan 
> 
> 
>On 04/01/2017 09:14 AM, Martin Morgan wrote: 
>> On 03/29/2017 03:55 PM, Henrik Bengtsson wrote: 
>>> Thanks. I have a few thoughts and questions in order to plan ahead: 
>>> 
>>>> - Our plan is to make a 'clean' transition from SVN to git, 
>>>> approximately one month after the next Bioconductor release. 
>>>> Developers or users will not have access to the SVN system after the 
>>>> date of transition. 
>>> 
>>> In order to preserve commit authorship, what's your plan for mapping 
>>> SVN username to Git 'user.email' and 'user.name'? The 'user.email' is 
>>> what GitHub uses to associate commits and contributions to GitHub 
>>> accounts. 
>> 
>> The svn administrators kept a comprehensive (but not complete) record 
>> of svn id, real user name, and contact email at the time the svn id 
>> was create. We use this to map between svn commits and git user name 
>> and email address. 
>> 
>> The information is not entirely consistent, with some fields for some 
>> records 'missing' (e.g., my record doesn't contain my real name) and 
>> of course out of date (e.g., my email address). 
>> 
>> Our intention is NOT to re-write history, but to map the information 
>> that we have to the git repositories. So my svn commits appear without 
>> a proper name, and with an outdated email address. Of course, new 
>> commits after the transition will contain whatever info git provides. 
>> 
>>> 
>>> BTW, using obsolete email addresses may prevent people from being 
>>> associated with those email addresses on GitHub and other online 
>>> services that require authentication of authorship claims (which go 
>>> out via those email addresses). 
>>> 
>>> When I SVN-to-Git exported my Bioconductor packages a few years ago, I 
>>> could handle manually because there were not too many contributors in 
>>> the SVN logs and I reached out to each of them and asked what email 
>>> addresses they would prefer to have in the Git commits. That approach 
>>> is obviously not feasible to automate for all Bioconductor packages. 
>>> Maybe this can be handled optionally by each package maintainers by 
>>> adding a .gitauthors file to the package root (or possibly via a 
>>> global Bioc one that everyone can commit to), e.g. 
>>> ```r 
>>> hb = Henrik Bengtsson <henr...@braju.com> 
>>> j...@foo.com = John Doe <j...@someone.org> 
>>> ``` 
>>> and then Bioc can use this mapping when exporting to Git? 
>>> 
>>> Finally, will people like me who already done the SVN-to-Git migration 
>>> be able to use that instead of the Bioc generated one? (I assume not, 
>>> but worth asking) 
>> 
>> No, we'll make a git snapshots of svn at the time of transition, and 
>> these git repositories will be the cannonical version -- developers 
>> will need to sync with these as they see fit. 
>> 
>> Martin 
>> 
>>> 
>>> Thanks, 
>>> 
>>> Henrik 
>>> 
>>> On Wed, Mar 29, 2017 at 9:51 AM, Turaga, Nitesh 
>>> <nitesh.tur...@roswellpark.org> wrote: 
>>>> Dear Bioconductor Developers, 
>>>> 
>>>> More news about the Git transition plan. We are coming close to our 
>>>> transition date and have made significant progress in getting our 

Re: [Bioc-devel] Git Transition Plan

[Turaga, Nitesh]

Hi Nathan,

We¹ve been working on a few things for the beta phase of the transition.

1. Getting the experiment data packages on board
2. Making sure manifests are in order along with their commit history.
3. Writing documentation for our new workflows for developers/maintainers.

The beta process has not started but, we expect it to be in play very
soon. 

New package submissions are still going through SVN. We will make sure to
send out an email/ post in the support group NEWS, and inform all the
users as soon as that happens.

You did not miss anything.

Best,

Nitesh




>Hi Nitesh and all,
>
>Can you give any status update on the SVN-to-git transition? I haven't
>seen any news since this exchange in March -- is the announced public
>beta in progress or ongoing? Are new package submissions going straight
>into git at this point?
>
>Or did I miss something?
>
>Thanks,
>
>Nathan
>
>
>On 04/01/2017 09:14 AM, Martin Morgan wrote:
>> On 03/29/2017 03:55 PM, Henrik Bengtsson wrote:
>>> Thanks.  I have a few thoughts and questions in order to plan ahead:
>>>
 - Our plan is to make a 'clean' transition from SVN to git,
 approximately one month after the next Bioconductor release.
 Developers or users will not have access to the SVN system after the
 date of transition.
>>>
>>> In order to preserve commit authorship, what's your plan for mapping
>>> SVN username to Git 'user.email' and 'user.name'?  The 'user.email' is
>>> what GitHub uses to associate commits and contributions to GitHub
>>> accounts.
>>
>> The svn administrators kept a comprehensive (but not complete) record
>> of svn id, real user name, and contact email at the time the svn id
>> was create. We use this to map between svn commits and git user name
>> and email address.
>>
>> The information is not entirely consistent, with some fields for some
>> records 'missing' (e.g., my record doesn't contain my real name) and
>> of course out of date (e.g., my email address).
>>
>> Our intention is NOT to re-write history, but to map the information
>> that we have to the git repositories. So my svn commits appear without
>> a proper name, and with an outdated email address. Of course, new
>> commits after the transition will contain whatever info git provides.
>>
>>>
>>> BTW, using obsolete email addresses may prevent people from being
>>> associated with those email addresses on GitHub and other online
>>> services that require authentication of authorship claims (which go
>>> out via those email addresses).
>>>
>>> When I SVN-to-Git exported my Bioconductor packages a few years ago, I
>>> could handle manually because there were not too many contributors in
>>> the SVN logs and I reached out to each of them and asked what email
>>> addresses they would prefer to have in the Git commits.  That approach
>>> is obviously not feasible to automate for all Bioconductor packages.
>>> Maybe this can be handled optionally by each package maintainers by
>>> adding a .gitauthors file to the package root (or possibly via a
>>> global Bioc one that everyone can commit to), e.g.
>>> ```r
>>> hb = Henrik Bengtsson 
>>> j...@foo.com = John Doe 
>>> ```
>>> and then Bioc can use this mapping when exporting to Git?
>>>
>>> Finally, will people like me who already done the SVN-to-Git migration
>>> be able to use that instead of the Bioc generated one? (I assume not,
>>> but worth asking)
>>
>> No, we'll make a git snapshots of svn at the time of transition, and
>> these git repositories will be the cannonical version -- developers
>> will need to sync with these as they see fit.
>>
>> Martin
>>
>>>
>>> Thanks,
>>>
>>> Henrik
>>>
>>> On Wed, Mar 29, 2017 at 9:51 AM, Turaga, Nitesh
>>>  wrote:
 Dear Bioconductor Developers,

 More news about the Git transition plan. We are coming close to our
 transition date and have made significant progress in getting our
 new server ready for the Bioconductor community.

 1. Overall plan:

 - Bioconductor hosts each package as a distinct repository at
 git.bioconductor.org. From
 Bioconductor's perspective, this is the canonical location. Nightly
 builds will be based on these repositories, release branches will be
 created in these repositories, etc. The naming convention for
 branches remains the same.

 - Developers clone or otherwise sync their code base to these
 repositories. Each developer will be able to push to and pull from
 (e.g., during branching and version bumps at package release) their
 git.bioconductor.org repository.
 Version bumps and new branches(during Bioconductor release) will be
 handled by the core team.

 - Developers are encouraged to host and develop their source code on
 Github or other git-based social platforms. This promotes community
 involvement, and empowers 

Re: [Bioc-devel] Git Transition Plan

[Nathan Sheffield]

Hi Nitesh and all,

Can you give any status update on the SVN-to-git transition? I haven't 
seen any news since this exchange in March -- is the announced public 
beta in progress or ongoing? Are new package submissions going straight 
into git at this point?


Or did I miss something?

Thanks,

Nathan


On 04/01/2017 09:14 AM, Martin Morgan wrote:

On 03/29/2017 03:55 PM, Henrik Bengtsson wrote:

Thanks.  I have a few thoughts and questions in order to plan ahead:

- Our plan is to make a 'clean' transition from SVN to git, 
approximately one month after the next Bioconductor release. 
Developers or users will not have access to the SVN system after the 
date of transition.


In order to preserve commit authorship, what's your plan for mapping
SVN username to Git 'user.email' and 'user.name'?  The 'user.email' is
what GitHub uses to associate commits and contributions to GitHub
accounts.


The svn administrators kept a comprehensive (but not complete) record 
of svn id, real user name, and contact email at the time the svn id 
was create. We use this to map between svn commits and git user name 
and email address.


The information is not entirely consistent, with some fields for some 
records 'missing' (e.g., my record doesn't contain my real name) and 
of course out of date (e.g., my email address).


Our intention is NOT to re-write history, but to map the information 
that we have to the git repositories. So my svn commits appear without 
a proper name, and with an outdated email address. Of course, new 
commits after the transition will contain whatever info git provides.




BTW, using obsolete email addresses may prevent people from being
associated with those email addresses on GitHub and other online
services that require authentication of authorship claims (which go
out via those email addresses).

When I SVN-to-Git exported my Bioconductor packages a few years ago, I
could handle manually because there were not too many contributors in
the SVN logs and I reached out to each of them and asked what email
addresses they would prefer to have in the Git commits.  That approach
is obviously not feasible to automate for all Bioconductor packages.
Maybe this can be handled optionally by each package maintainers by
adding a .gitauthors file to the package root (or possibly via a
global Bioc one that everyone can commit to), e.g.
```r
hb = Henrik Bengtsson 
j...@foo.com = John Doe 
```
and then Bioc can use this mapping when exporting to Git?

Finally, will people like me who already done the SVN-to-Git migration
be able to use that instead of the Bioc generated one? (I assume not,
but worth asking)


No, we'll make a git snapshots of svn at the time of transition, and 
these git repositories will be the cannonical version -- developers 
will need to sync with these as they see fit.


Martin



Thanks,

Henrik

On Wed, Mar 29, 2017 at 9:51 AM, Turaga, Nitesh
 wrote:

Dear Bioconductor Developers,

More news about the Git transition plan. We are coming close to our 
transition date and have made significant progress in getting our 
new server ready for the Bioconductor community.


1. Overall plan:

- Bioconductor hosts each package as a distinct repository at 
git.bioconductor.org. From 
Bioconductor's perspective, this is the canonical location. Nightly 
builds will be based on these repositories, release branches will be 
created in these repositories, etc. The naming convention for 
branches remains the same.


- Developers clone or otherwise sync their code base to these 
repositories. Each developer will be able to push to and pull from 
(e.g., during branching and version bumps at package release) their 
git.bioconductor.org repository. 
Version bumps and new branches(during Bioconductor release) will be 
handled by the core team.


- Developers are encouraged to host and develop their source code on 
Github or other git-based social platforms. This promotes community 
involvement, and empowers developers to adopt best practices related 
to issue tracking, continuous integration, bug fixes, pull requests 
from their community, etc.


- All bioconductor infrastructure code will also be available on 
Github, through our organization 
page(https://github.com/Bioconductor). Community members are 
encouraged to send us pull requests for all our public repositories.


2. Timing:

- Our plan is to make a 'clean' transition from SVN to git, 
approximately one month after the next Bioconductor release. 
Developers or users will not have access to the SVN system after the 
date of transition.


3. Repositories:

- The git repositories will be derived from a 'snapshot' of the 
latest SVN repository at the time of the transition. After the date 
of transition, further commits to SVN will not be reflected in the 
new git repositories.


- Each repository will capture the full SVN 

Re: [Bioc-devel] Git Transition Plan

[Martin Morgan]

On 03/29/2017 03:55 PM, Henrik Bengtsson wrote:

Thanks.  I have a few thoughts and questions in order to plan ahead:


- Our plan is to make a 'clean' transition from SVN to git, approximately one 
month after the next Bioconductor release. Developers or users will not have 
access to the SVN system after the date of transition.


In order to preserve commit authorship, what's your plan for mapping
SVN username to Git 'user.email' and 'user.name'?  The 'user.email' is
what GitHub uses to associate commits and contributions to GitHub
accounts.


The svn administrators kept a comprehensive (but not complete) record of 
svn id, real user name, and contact email at the time the svn id was 
create. We use this to map between svn commits and git user name and 
email address.


The information is not entirely consistent, with some fields for some 
records 'missing' (e.g., my record doesn't contain my real name) and of 
course out of date (e.g., my email address).


Our intention is NOT to re-write history, but to map the information 
that we have to the git repositories. So my svn commits appear without a 
proper name, and with an outdated email address. Of course, new commits 
after the transition will contain whatever info git provides.




BTW, using obsolete email addresses may prevent people from being
associated with those email addresses on GitHub and other online
services that require authentication of authorship claims (which go
out via those email addresses).

When I SVN-to-Git exported my Bioconductor packages a few years ago, I
could handle manually because there were not too many contributors in
the SVN logs and I reached out to each of them and asked what email
addresses they would prefer to have in the Git commits.  That approach
is obviously not feasible to automate for all Bioconductor packages.
Maybe this can be handled optionally by each package maintainers by
adding a .gitauthors file to the package root (or possibly via a
global Bioc one that everyone can commit to), e.g.
```r
hb = Henrik Bengtsson 
j...@foo.com = John Doe 
```
and then Bioc can use this mapping when exporting to Git?

Finally, will people like me who already done the SVN-to-Git migration
be able to use that instead of the Bioc generated one? (I assume not,
but worth asking)


No, we'll make a git snapshots of svn at the time of transition, and 
these git repositories will be the cannonical version -- developers will 
need to sync with these as they see fit.


Martin



Thanks,

Henrik

On Wed, Mar 29, 2017 at 9:51 AM, Turaga, Nitesh
 wrote:

Dear Bioconductor Developers,

More news about the Git transition plan. We are coming close to our transition 
date and have made significant progress in getting our new server ready for the 
Bioconductor community.

1. Overall plan:

- Bioconductor hosts each package as a distinct repository at 
git.bioconductor.org. From Bioconductor's 
perspective, this is the canonical location. Nightly builds will be based on these 
repositories, release branches will be created in these repositories, etc. The naming 
convention for branches remains the same.

- Developers clone or otherwise sync their code base to these repositories. Each 
developer will be able to push to and pull from (e.g., during branching and version 
bumps at package release) their 
git.bioconductor.org repository. Version bumps 
and new branches(during Bioconductor release) will be handled by the core team.

- Developers are encouraged to host and develop their source code on Github or 
other git-based social platforms. This promotes community involvement, and 
empowers developers to adopt best practices related to issue tracking, 
continuous integration, bug fixes, pull requests from their community, etc.

- All bioconductor infrastructure code will also be available on Github, 
through our organization page(https://github.com/Bioconductor). Community 
members are encouraged to send us pull requests for all our public repositories.

2. Timing:

- Our plan is to make a 'clean' transition from SVN to git, approximately one 
month after the next Bioconductor release. Developers or users will not have 
access to the SVN system after the date of transition.

3. Repositories:

- The git repositories will be derived from a 'snapshot' of the latest SVN 
repository at the time of the transition. After the date of transition, further 
commits to SVN will not be reflected in the new git repositories.

- Each repository will capture the full SVN commit history of the package. 
Releases will be included as branches in the repository.

- We anticipate that software and experiment data packages will be hosted at 
git.bioconductor.org. Experiment data packages 
will use git-lfs to manage large data.

- The Github Bioconductor-mirror repository will be removed.

4. Test phases:

- 

Re: [Bioc-devel] Git Transition Plan

[Henrik Bengtsson]

Thanks.  I have a few thoughts and questions in order to plan ahead:

> - Our plan is to make a 'clean' transition from SVN to git, approximately one 
> month after the next Bioconductor release. Developers or users will not have 
> access to the SVN system after the date of transition.

In order to preserve commit authorship, what's your plan for mapping
SVN username to Git 'user.email' and 'user.name'?  The 'user.email' is
what GitHub uses to associate commits and contributions to GitHub
accounts.

BTW, using obsolete email addresses may prevent people from being
associated with those email addresses on GitHub and other online
services that require authentication of authorship claims (which go
out via those email addresses).

When I SVN-to-Git exported my Bioconductor packages a few years ago, I
could handle manually because there were not too many contributors in
the SVN logs and I reached out to each of them and asked what email
addresses they would prefer to have in the Git commits.  That approach
is obviously not feasible to automate for all Bioconductor packages.
Maybe this can be handled optionally by each package maintainers by
adding a .gitauthors file to the package root (or possibly via a
global Bioc one that everyone can commit to), e.g.
```r
hb = Henrik Bengtsson 
j...@foo.com = John Doe 
```
and then Bioc can use this mapping when exporting to Git?

Finally, will people like me who already done the SVN-to-Git migration
be able to use that instead of the Bioc generated one? (I assume not,
but worth asking)

Thanks,

Henrik

On Wed, Mar 29, 2017 at 9:51 AM, Turaga, Nitesh
 wrote:
> Dear Bioconductor Developers,
>
> More news about the Git transition plan. We are coming close to our 
> transition date and have made significant progress in getting our new server 
> ready for the Bioconductor community.
>
> 1. Overall plan:
>
> - Bioconductor hosts each package as a distinct repository at 
> git.bioconductor.org. From Bioconductor's 
> perspective, this is the canonical location. Nightly builds will be based on 
> these repositories, release branches will be created in these repositories, 
> etc. The naming convention for branches remains the same.
>
> - Developers clone or otherwise sync their code base to these repositories. 
> Each developer will be able to push to and pull from (e.g., during branching 
> and version bumps at package release) their 
> git.bioconductor.org repository. Version bumps 
> and new branches(during Bioconductor release) will be handled by the core 
> team.
>
> - Developers are encouraged to host and develop their source code on Github 
> or other git-based social platforms. This promotes community involvement, and 
> empowers developers to adopt best practices related to issue tracking, 
> continuous integration, bug fixes, pull requests from their community, etc.
>
> - All bioconductor infrastructure code will also be available on Github, 
> through our organization page(https://github.com/Bioconductor). Community 
> members are encouraged to send us pull requests for all our public 
> repositories.
>
> 2. Timing:
>
> - Our plan is to make a 'clean' transition from SVN to git, approximately one 
> month after the next Bioconductor release. Developers or users will not have 
> access to the SVN system after the date of transition.
>
> 3. Repositories:
>
> - The git repositories will be derived from a 'snapshot' of the latest SVN 
> repository at the time of the transition. After the date of transition, 
> further commits to SVN will not be reflected in the new git repositories.
>
> - Each repository will capture the full SVN commit history of the package. 
> Releases will be included as branches in the repository.
>
> - We anticipate that software and experiment data packages will be hosted at 
> git.bioconductor.org. Experiment data packages 
> will use git-lfs to manage large data.
>
> - The Github Bioconductor-mirror repository will be removed.
>
> 4. Test phases:
>
> - We anticipate two test periods.
>
> - A private 'alpha' test phase is starting shortly. Alpha testers will 
> initially have pull access to their repository, and will follow tentative 
> work flows for gaining push access, for maintaining clones of their packages, 
> for committing to the git.bioconductor.org 
> repositories, and for pulling from (e.g., version bumps and release branches) 
> the repositories.
>
> - A public 'beta' test phase will start after the next Bioconductor release, 
> and last for approximately four weeks. Beta testers will have pull access to 
> all Bioconductor package repositories, and push access to packages that they 
> maintain. Commits will be visible, but NOT synced with SVN or incorporated 
> into nightly builds.
>
> - An essential activity during test periods is to verify that the