Subject: Re: [boost] Refactoring tools/regression and tools/release into separate sub-modules
From: Daniel James (dnljms_at_[hidden])
Date: 2014-02-11 18:00:08
On 11 February 2014 21:15, Ahmed Charles <acharles_at_[hidden]> wrote:
>> From: dnljms_at_[hidden]
>> Looks good, would it be possible to describe the method you used? As
>> others might need to do something similar.
> I used filter-branch --subdirectory-filter, basically. The only interesting thing other than that is that I also added a .gitattributes file to each commit using a --tree-filter so that checkout's of the history work well with file endings (or at least as well as they do currently).
Thanks for the info, adding gitattributes is a good idea.
> Note that I didn't migrate tags over, mostly because I deleted the local copy before I realized that I hadn't pushed them. Anyways, if you want tags as well, I can just redo the commands tonight and upload new versions of the repositories.
I don't think it really matters for these modules, unless someone disagrees.
>> I think the next steps are:
>> 1. Transfer the modules into 'boostorg', I think you can do that
>> through github, although you might have to be added to a team if you
>> don't have any write permissions in the organisation.
> I don't have perms on any team, so I can't do that.
I added you to a new 'regression' team, so you should be able to now.
I'll downgrade it to write access once you're done.
>> 2. Change the regression testers to use the regression module. We need
>> to do that before updating the super project as they download directly
>> from it.
> Ah, who do I have to coordinate with to get this done?
I think you just need to send an email to the testing list:
> In terms of steps:
> 1. Make sure the content is right (mostly whether tags are desired or not).
> 2. Move the repositories into the boostorg organisation (I can do this if I get permissions and signoff on step 1).
> 3. Figure out who needs to change the regression testers for this to work.
> 4. Determine if the superproject content should simply be deleted or if it should contain a submodule reference. (Who do I talk to about regression or release and whether they should be a submodule or not?)
Beman and Marshall are the release people. If they have a problem with
removing it from the tree they should speak up now. For the regression
module, see what people on the testing list say.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk