www-site-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sebb <seb...@gmail.com>
Subject Re: svn commit: r1786095 - /infrastructure/site/trunk/content/dev/infra-contact.mdtext
Date Thu, 09 Mar 2017 11:59:14 GMT
On 9 March 2017 at 11:20, Greg Stein <gstein@gmail.com> wrote:
> On Thu, Mar 9, 2017 at 5:08 AM, sebb <sebbaz@gmail.com> wrote:
>>
>> On 9 March 2017 at 03:26,  <gstein@apache.org> wrote:
>> > Author: gstein
>> > Date: Thu Mar  9 03:26:40 2017
>> > New Revision: 1786095
>> >
>> > URL: http://svn.apache.org/viewvc?rev=1786095&view=rev
>> > Log:
>> > * svn2git is now self-serve. add some HOWTO notes.
>> > * adjust contact address for Infra
>> >
>> > Modified:
>> >     infrastructure/site/trunk/content/dev/infra-contact.mdtext
>
>>...
>>
>> > +| Migrate your project's **SVN to Git** | n/a | Please use
>> > [reporeq](https://reporeq.apache.org) to create your intended Git repo(s).
>> > Run svn2git locally using this [authors
>> > file](https://git-wip-us.apache.org/authors.txt) and push once the
>> > conversion result is confirmed. File an INFRA ticket to mark your SVN
>> > repository readonly. Optionally, file a ticket to temporarily disable commit
>> > emails for when you push your converted clone. |
>>
>> Would it not be safer to mark SVN read-only *before* doing the conversion?
>>
>> [If users want to experiment with svn2git first they can use a
>> temporary local git repo.]
>
>
> "Safer" ? ... nah. Nothing is getting broken.
>
> Could an svn commit arrive, while a contributor is working through the
> conversion process, and verifying its result? Sure. But they can' just
> re-run their conversion.

They can, if they notice that SVN has been updated.

But if they don't notice, or the commit mail gets delayed or lost and
they don't recheck the SVN version manually, then the Git repo is
broken.

This process used to be undertaken by infra staff who should be well
aware of the potential for lost updates.

The average committer who now has to do the task may not have
considered such possibilities.

That is why it is safer to mark SVN read-only first.

> And yes: the intent is that a PMC/community uses a "temporary local git
> repo". Running through conversion tests on the live new/empty Git repo is
> just... nonsensical.

I did not mean to suggest that the user would push the result of
running svn2git until they were happy with it.

> Your thoughts on wording is most welcome.

I thought I already commented ...

> This is kind of a big change
> around what Infra will help people with on svn->git. A good description of
> the new position would be welcome!

No idea what you mean by that.

> Thx,
> -g
>

Mime
View raw message