[Conversion] Ready to turn off write access to svn?

Beman Dawes

19 Nov 2013 19 Nov '13

1:21 p.m.

According to the calendar (http://www.boost.org/community/index.html), it is time to turn off write access to the Subversion repo. Are there any showstoppers? --Beman

Show replies by date

Paul A. Bristow

19 Nov 19 Nov

3:18 p.m.

...

-----Original Message----- From: Boost [mailto:boost-bounces@lists.boost.org] On Behalf Of Beman Dawes Sent: Tuesday, November 19, 2013 1:21 PM To: Boost Developers List Subject: [boost] [Conversion] Ready to turn off write access to svn?

According to the calendar (http://www.boost.org/community/index.html), it is time to turn off write access to the Subversion repo.

Are there any showstoppers?

Will this mean boost-sandbox as well? Some of us planning converting sandbox items to GIT might find it easier if we could use Subversion until we are fully GITerated? Paul --- Paul A. Bristow, Prizet Farmhouse, Kendal LA8 8AB UK +44 1539 561830 07714330204 pbristow@hetp.u-net.com

Beman Dawes

20 Nov 20 Nov

1:04 p.m.

On Tue, Nov 19, 2013 at 10:18 AM, Paul A. Bristow wrote:

...

...
-----Original Message----- From: Boost [mailto:boost-bounces@lists.boost.org] On Behalf Of Beman Dawes Sent: Tuesday, November 19, 2013 1:21 PM To: Boost Developers List Subject: [boost] [Conversion] Ready to turn off write access to svn?

According to the calendar (http://www.boost.org/community/index.html), it is time to turn off write access to the Subversion repo.

Are there any showstoppers?

Will this mean boost-sandbox as well?

Yes. My understanding of the conversion process is that it treats the svn repo as a whole, including the sandbox.

...

Some of us planning converting sandbox items to GIT might find it easier if we could use Subversion until we are fully GITerated?

Have you read "Moving Sandbox Libraries to GitHub"? https://svn.boost.org/trac/boost/wiki/ExtractSandbox Give it a try. Remember that it is trivial to delete a repo from your GitHub account, so you can do the conversion and play around with the resulting GitHub repo to develop confidence. The just delete the GitHub repo and to the conversion for real. HTH, --Beman

Vicente J. Botet Escriba

19 Nov 19 Nov

6:41 p.m.

Le 19/11/13 14:21, Beman Dawes a écrit :

...

According to the calendar (http://www.boost.org/community/index.html), it is time to turn off write access to the Subversion repo.

Are there any showstoppers?

Hi, I've heard about some documentation that should be written. I'm not a Git expert and I would like some help. Where can I find it? Best, Vicente

Beman Dawes

8:13 p.m.

On Tue, Nov 19, 2013 at 1:41 PM, Vicente J. Botet Escriba < vicente.botet@wanadoo.fr> wrote:

...

I've heard about some documentation that should be written. I'm not a Git expert and I would like some help. Where can I find it?

Getting started with Git - https://svn.boost.org/trac/boost/wiki/Git/GitHome Getting Started with GitHub - https://svn.boost.org/trac/boost/wiki/StartGitHub Getting Started with Modular Boost - https://svn.boost.org/trac/boost/wiki/TryModBoost Getting Started with Modular Boost Library Development Workflow - https://svn.boost.org/trac/boost/wiki/StartModWorkflow These will be updated this week as the final details of the conversion freeze into place. --Beman

Ion Gaztañaga

9:44 p.m.

El 19/11/2013 21:13, Beman Dawes escribió:

...

Getting started with Git - https://svn.boost.org/trac/boost/wiki/Git/GitHome Getting Started with GitHub - https://svn.boost.org/trac/boost/wiki/StartGitHub Getting Started with Modular Boost - https://svn.boost.org/trac/boost/wiki/TryModBoost Getting Started with Modular Boost Library Development Workflow - https://svn.boost.org/trac/boost/wiki/StartModWorkflow

These will be updated this week as the final details of the conversion freeze into place.

Many thanks Beman. And thanks to all working on this. I must admit that I'm a bit lazy to tweak the way we've worked all these years and I haven't followed the conversion process, but I think it's time for a change ;-) Best, Ion

Edward Diener

10:52 p.m.

On 11/19/2013 3:13 PM, Beman Dawes wrote:

...

On Tue, Nov 19, 2013 at 1:41 PM, Vicente J. Botet Escriba < vicente.botet@wanadoo.fr> wrote:

...
I've heard about some documentation that should be written. I'm not a Git expert and I would like some help. Where can I find it?

Getting started with Git - https://svn.boost.org/trac/boost/wiki/Git/GitHome Getting Started with GitHub - https://svn.boost.org/trac/boost/wiki/StartGitHub Getting Started with Modular Boost - https://svn.boost.org/trac/boost/wiki/TryModBoost Getting Started with Modular Boost Library Development Workflow - https://svn.boost.org/trac/boost/wiki/StartModWorkflow

These will be updated this week as the final details of the conversion freeze into place.

Many thanks for all your work keeping the documentation for Git and modular Boost up to date. I had been making some changes to trunk, not release, but I can wait until after the Git transition to complete my work. When Boost decides to officially make SVN read-only could you post exactly when this is going to be done so that I will not try to update SVN after that date/time. Also once the transition to Git is complete and we should all be using Git can you post when that occurs, so I can transiiton to Git. Needless to say I hope all the Boost Git experts will try to answer questions about the transition if a Git novice like me runs into any problems.

Beman Dawes

20 Nov 20 Nov

1:41 p.m.

On Tue, Nov 19, 2013 at 5:52 PM, Edward Diener wrote:

...

... I had been making some changes to trunk, not release, but I can wait until after the Git transition to complete my work. When Boost decides to officially make SVN read-only could you post exactly when this is going to be done so that I will not try to update SVN after that date/time.

The plan is to add an SVN commit hook that prevents changes. There should be an associated message pointing to boostorg on GitHub.

...

Also once the transition to Git is complete and we should all be using Git can you post when that occurs, so I can transiiton to Git.

Right. The key steps are: * Closing svn for commits. Of course the svn repo will be preserved for historical purposes. * Shutting down the conversion process, so that the boostorg GitHub repos are no longer being overwritten. * Opening the boostorg GitHub repos for use. Each of these steps is preceded by some inspection and testing, and will be announced on the Boost list.

...

Needless to say I hope all the Boost Git experts will try to answer questions about the transition if a Git novice like me runs into any problems.

For sure:-! --Beman

Jim Bell

4:54 p.m.

On 2013-11-20 7:41 AM, Beman Dawes wrote:

...

... Right.

The key steps are:

* Closing svn for commits. Of course the svn repo will be preserved for historical purposes. * Shutting down the conversion process, so that the boostorg GitHub repos are no longer being overwritten. * Opening the boostorg GitHub repos for use.

Each of these steps is preceded by some inspection and testing, and will be announced on the Boost list.

What impact does this have on regression testers? Do the regression scripts handle everything, or do testers need to take some action? (E.g., installing git.) I suggest posting to the boost-testing list to give a the testers what they need to know (and a heads-up). I don't follow this list closely, and just happened to come across this post. Thanks, -Jim

Steven Watanabe

5:22 p.m.

AMDG On 11/20/2013 08:54 AM, Jim Bell wrote:

...

On 2013-11-20 7:41 AM, Beman Dawes wrote:

...
... Right.

The key steps are:

* Closing svn for commits. Of course the svn repo will be preserved for historical purposes. * Shutting down the conversion process, so that the boostorg GitHub repos are no longer being overwritten. * Opening the boostorg GitHub repos for use.

Each of these steps is preceded by some inspection and testing, and will be announced on the Boost list.

What impact does this have on regression testers? Do the regression scripts handle everything, or do testers need to take some action? (E.g., installing git.)

The regression scripts don't necessarily need git. There's an option to download a tarball from beta.boost.org, instead. However, the regression scripts don't currently have any support for git at all.

...

I suggest posting to the boost-testing list to give a the testers what they need to know (and a heads-up). I don't follow this list closely, and just happened to come across this post.

In Christ, Steven Watanabe

Rene Rivera

21 Nov 21 Nov

4:04 a.m.

On Wed, Nov 20, 2013 at 11:22 AM, Steven Watanabe wrote:

...

AMDG

On 11/20/2013 08:54 AM, Jim Bell wrote:

...
On 2013-11-20 7:41 AM, Beman Dawes wrote:

...
... Right.

The key steps are:

* Closing svn for commits. Of course the svn repo will be preserved for historical purposes. * Shutting down the conversion process, so that the boostorg GitHub

repos

...
...
are no longer being overwritten. * Opening the boostorg GitHub repos for use.

Each of these steps is preceded by some inspection and testing, and will be announced on the Boost list.

What impact does this have on regression testers? Do the regression scripts handle everything, or do testers need to take some action? (E.g., installing git.)

Testing will stop showing changes as the work to switch the testing to use github is not complete. And I don't have the time right now to do that work. And will not have the time for a few weeks.

...

The regression scripts don't necessarily need git. There's an option to download a tarball from beta.boost.org, instead. However, the regression scripts don't currently have any support for git at all.

They do. When I first looked at github I really wanted to use that tar-ball option. But those archives do not work at all with the external/symlink (or whatever is the name of the subproject references that I can't remember at the moment). ..Now I need to get back to work on my Alpha deadline. -- -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo

Jim Bell

7:54 p.m.

On 2013-11-20 10:04 PM, Rene Rivera wrote:

...

...

...
...
What impact does this have on regression testers? ...

Testing will stop showing changes as the work to switch the testing to use github is not complete. And I don't have the time right now to do that work. And will not have the time for a few weeks.

...
The regression scripts don't necessarily need git. ...

They do. ....

Doesn't this mean that, when write access to svn is switched off, we also need to tell regression testers to stop until further notice?

Vicente J. Botet Escriba

10:03 p.m.

Le 21/11/13 20:54, Jim Bell a écrit :

...

Doesn't this mean that, when write access to svn is switched off, we also need to tell regression testers to stop until further notice?

It would be good if all tester have been run at least once when svn had been frozen. Best, Vicente

Beman Dawes

22 Nov 22 Nov

1:31 p.m.

On Thu, Nov 21, 2013 at 5:03 PM, Vicente J. Botet Escriba < vicente.botet@wanadoo.fr> wrote:

...

Le 21/11/13 20:54, Jim Bell a écrit :

...
Doesn't this mean that, when write access to svn is switched off, we also need to tell regression testers to stop until further notice?

It would be good if all tester have been run at least once when svn had

been frozen.

Certainly we will want to let the nightly testers cycle testing against svn. And part of the integrity tests that will be run after svn freezes is to verify that every test for every library produces the same result against both the svn and git repos. --Beman

Beman Dawes

1:22 p.m.

On Thu, Nov 21, 2013 at 2:54 PM, Jim Bell wrote:

...

On 2013-11-20 10:04 PM, Rene Rivera wrote:

...
...

...
What impact does this have on regression testers? ...

...
Testing will stop showing changes as the work to switch the testing to use github is not complete. And I don't have the time right now to do that work. And will not have the time for a few weeks.

The regression scripts don't necessarily need git. ...

...
They do. ....

Doesn't this mean that, when write access to svn is switched off, we also need to tell regression testers to stop until further notice?

It isn't clear to me yet exactly what testers will have to do, if anything. But we do need to let regression testers know as the details firm. --Beman

Rene Rivera

3:45 p.m.

On Wed, Nov 20, 2013 at 10:04 PM, Rene Rivera wrote:

...

On Wed, Nov 20, 2013 at 11:22 AM, Steven Watanabe wrote:

...
AMDG

On 11/20/2013 08:54 AM, Jim Bell wrote:

...
On 2013-11-20 7:41 AM, Beman Dawes wrote:

...
... Right.

The key steps are:

* Closing svn for commits. Of course the svn repo will be preserved for historical purposes. * Shutting down the conversion process, so that the boostorg GitHub

repos

...
...
are no longer being overwritten. * Opening the boostorg GitHub repos for use.

Each of these steps is preceded by some inspection and testing, and will be announced on the Boost list.

What impact does this have on regression testers? Do the regression scripts handle everything, or do testers need to take some action? (E.g., installing git.)

Testing will stop showing changes as the work to switch the testing to use github is not complete. And I don't have the time right now to do that work. And will not have the time for a few weeks.

I briefly checked, while waiting for a contractor to be late, as to the progress of conversion of the testing scripts. And it's not bad.. I was almost done with the conversion. Unfortunately I just tested fetching the main boost git repo and the fetch failed for dulwich doesn't directly recognize the new relative subrepo paths we are using. So it will take some minor tweaks to get that part working again. It's only a small amount of work left at this point. The regression scripts don't necessarily need git.

...

...
There's an option to download a tarball from beta.boost.org, instead. However, the regression scripts don't currently have any support for git at all.

They do. When I first looked at github I really wanted to use that tar-ball option. But those archives do not work at all with the external/symlink (or whatever is the name of the subproject references that I can't remember at the moment).

I should have pointed out that I'm using dulwich. So tester will not need to install git per se. As we will get dulwich directly for the git part.

...

..Now I need to get back to work on my Alpha deadline.

Alpha was cleared last night :-) -- -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo

Rene Rivera

24 Nov 24 Nov

5:27 a.m.

This is last minute.. Attached are changes I have locally to the regression testing scripts that have my latest work. I can't really check them in as they are untested and hence I don't know if they would break the current testing (which we need for a little bit longer). If someone else can continue the work that would be great. Note, all the files are in root/tools/regression/src. On Fri, Nov 22, 2013 at 9:45 AM, Rene Rivera wrote:

...

On Wed, Nov 20, 2013 at 10:04 PM, Rene Rivera wrote:

...
On Wed, Nov 20, 2013 at 11:22 AM, Steven Watanabe wrote:

...
AMDG

On 11/20/2013 08:54 AM, Jim Bell wrote:

...
On 2013-11-20 7:41 AM, Beman Dawes wrote:

...
... Right.

The key steps are:

* Closing svn for commits. Of course the svn repo will be preserved

for

...
...
historical purposes. * Shutting down the conversion process, so that the boostorg GitHub repos are no longer being overwritten. * Opening the boostorg GitHub repos for use.

Each of these steps is preceded by some inspection and testing, and will be announced on the Boost list.

What impact does this have on regression testers? Do the regression scripts handle everything, or do testers need to take some action? (E.g., installing git.)

Testing will stop showing changes as the work to switch the testing to use github is not complete. And I don't have the time right now to do that work. And will not have the time for a few weeks.

I briefly checked, while waiting for a contractor to be late, as to the progress of conversion of the testing scripts. And it's not bad.. I was almost done with the conversion. Unfortunately I just tested fetching the main boost git repo and the fetch failed for dulwich doesn't directly recognize the new relative subrepo paths we are using. So it will take some minor tweaks to get that part working again. It's only a small amount of work left at this point.

The regression scripts don't necessarily need git.

...
...
There's an option to download a tarball from beta.boost.org, instead. However, the regression scripts don't currently have any support for git at all.

They do. When I first looked at github I really wanted to use that tar-ball option. But those archives do not work at all with the external/symlink (or whatever is the name of the subproject references that I can't remember at the moment).

I should have pointed out that I'm using dulwich. So tester will not need to install git per se. As we will get dulwich directly for the git part.

...
..Now I need to get back to work on my Alpha deadline.

Alpha was cleared last night :-)

-- -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo

-- -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo

4169

Age (days ago)

4174

Last active (days ago)

List overview

Download

16 comments

8 participants

participants (8)

Beman Dawes
Edward Diener
Ion Gaztañaga
Jim Bell
Paul A. Bristow
Rene Rivera
Steven Watanabe
Vicente J. Botet Escriba