Re: Bugzilla migration tool user accounts

2017-11-30 Thread philip . chimento
On Thu, Nov 30, 2017 at 9:51 AM Nicolas Dufresne 
wrote:

> Le jeudi 30 novembre 2017 à 12:54 +0100, Alexandre Franke a écrit :
> > On Thu, Nov 30, 2017 at 6:43 AM,   wrote:
> > > Bastien, do you also mean to imply that the wish to be subscribed
> > > to >2000
> > > issues means that receiving >2000 mails won't be a problem?
> >
> > It will be a problem, but not as bad as losing subscription on those
> > issues. It would still be very much worth looking for a way to
> > temporarily disable notifications to avoid that flood. I can also
> > imagine that sending that many email out at once can raise some flags
> > and make GNOME look bad in the eye of some email providers.
>
> This was an issue when some project tried to migrate to Phabricator.
> The server end up stalled for days, spamming everyone. We need to be
> careful.
>

Carlos thought it would be possible to just stop all email while a
migration was running; it sounds like that's what we should plan to do.

Regards,
Philip C
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
https://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: Bugzilla migration tool user accounts

2017-11-30 Thread philip . chimento
On Thu, Nov 30, 2017 at 5:08 AM Germán Poo-Caamaño  wrote:

> On Tue, 2017-11-28 at 04:21 +, philip.chime...@gmail.com wrote:
> > Hi list,
> >
> > This is about the migration tool for Bugzilla bug reports to GitLab
> > issues.
> > See, for background, issues [1] and [2]. Currently, the migration
> > tool
> > posts all the migrated issues and their associated comments as one
> > single
> > user; either the maintainer who runs the script, or a special
> > 'bugzilla-migration' user.
> >
> > I'd like to open the discussion, as suggested in those two issues,
> > about
> > posting comments as the actual comment author (if their email can be
> > matched to a GitLab user.) So, instead of comments coming from
> > 'bugzilla-migration' or the maintainer who runs the script, if I had
> > commented on a Bugzilla bug that was migrated then the comment on
> > GitLab
> > would actually be coming from the @ptomato account, and we wouldn't
> > need
> > all the extra quoting.
> >
> > On the pro side,
> > 1. the flow of the comments is more natural and readable that way.
> > 2. it ensures that existing subscribers to a Bugzilla bug remain
> > subscribed
> > after the migration.
> > 3. it associates people's past contributions to GNOME with their
> > current
> > bugtracker accounts.
> >
> > On the con side,
> > 4. "impersonating" a GitLab user, even to post content that they had
> > already written elsewhere, could be uncomfortable to some people.
> > 5. it would send a lot of GitLab email which would be annoying.
> > 6. it would require giving the 'bugzilla-migration' account admin
> > access,
> > which means that access tokens wouldn't be given out so easily to
> > maintainers in order to run the migration script.
>
> Hi Philip,
>
> Would some authorship metadata be kept?
>
> If I understand correctly, a major con is that we would not able to
> discriminate who said what, nor who replied. Was the comment made by a
> newcomer? a maintainer? a developer?
>
> Reports and comments are not equally weighed. For example, Bryan Clark
> does not contribute to GNOME anymore, but his comments on Evince are
> weighed very high, because he was the original designer of Evince.
>
> Another (minor) con is that we would not be able to search by
> user/reporter/developer. Sometimes, I do not remember exactly the bug
> or text, but I do remember who participated and I can narrow the search
> faster.
>
> In practical terms, not migrating the user/emails sounds like many bugs
> should be re-triaged.
>
> Please, correct me if I am wrong with my perception.
>

Hi Germán,

Even if we do not post the comment impersonating the original author's
account, it will still be clear who wrote what. We will still need to make
that clear anyway for Bugzilla accounts who don't have a GitLab account.

Check [1] (tagging the original author) and [2] (not tagging the original
author; note these links are not intended to be permanent) for examples of
how the migration tool does it currently.

Otherwise you are correct, not impersonating the user accounts will make it
hard to search for who posted a comment.

Regards,
Philip C

[1] https://gitlab-test.gnome.org/ptomato/gjs-test/issues/75
[2] https://gitlab-test.gnome.org/ptomato/gjs-test2/issues/75
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
https://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: Bugzilla migration tool user accounts

2017-11-30 Thread Nicolas Dufresne
Le jeudi 30 novembre 2017 à 12:54 +0100, Alexandre Franke a écrit :
> On Thu, Nov 30, 2017 at 6:43 AM,   wrote:
> > Bastien, do you also mean to imply that the wish to be subscribed
> > to >2000
> > issues means that receiving >2000 mails won't be a problem?
> 
> It will be a problem, but not as bad as losing subscription on those
> issues. It would still be very much worth looking for a way to
> temporarily disable notifications to avoid that flood. I can also
> imagine that sending that many email out at once can raise some flags
> and make GNOME look bad in the eye of some email providers.

This was an issue when some project tried to migrate to Phabricator.
The server end up stalled for days, spamming everyone. We need to be
careful.

Nicolas

signature.asc
Description: This is a digitally signed message part
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
https://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: Bugzilla migration tool user accounts

2017-11-30 Thread Germán Poo-Caamaño
On Tue, 2017-11-28 at 04:21 +, philip.chime...@gmail.com wrote:
> Hi list,
> 
> This is about the migration tool for Bugzilla bug reports to GitLab
> issues.
> See, for background, issues [1] and [2]. Currently, the migration
> tool
> posts all the migrated issues and their associated comments as one
> single
> user; either the maintainer who runs the script, or a special
> 'bugzilla-migration' user.
> 
> I'd like to open the discussion, as suggested in those two issues,
> about
> posting comments as the actual comment author (if their email can be
> matched to a GitLab user.) So, instead of comments coming from
> 'bugzilla-migration' or the maintainer who runs the script, if I had
> commented on a Bugzilla bug that was migrated then the comment on
> GitLab
> would actually be coming from the @ptomato account, and we wouldn't
> need
> all the extra quoting.
> 
> On the pro side,
> 1. the flow of the comments is more natural and readable that way.
> 2. it ensures that existing subscribers to a Bugzilla bug remain
> subscribed
> after the migration.
> 3. it associates people's past contributions to GNOME with their
> current
> bugtracker accounts.
> 
> On the con side,
> 4. "impersonating" a GitLab user, even to post content that they had
> already written elsewhere, could be uncomfortable to some people.
> 5. it would send a lot of GitLab email which would be annoying.
> 6. it would require giving the 'bugzilla-migration' account admin
> access,
> which means that access tokens wouldn't be given out so easily to
> maintainers in order to run the migration script.

Hi Philip,

Would some authorship metadata be kept?

If I understand correctly, a major con is that we would not able to
discriminate who said what, nor who replied. Was the comment made by a
newcomer? a maintainer? a developer?

Reports and comments are not equally weighed. For example, Bryan Clark
does not contribute to GNOME anymore, but his comments on Evince are
weighed very high, because he was the original designer of Evince.

Another (minor) con is that we would not be able to search by
user/reporter/developer. Sometimes, I do not remember exactly the bug
or text, but I do remember who participated and I can narrow the search
faster.

In practical terms, not migrating the user/emails sounds like many bugs
should be re-triaged.

Please, correct me if I am wrong with my perception.

-- 
Germán Poo-Caamaño
http://calcifer.org/

signature.asc
Description: This is a digitally signed message part
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
https://mail.gnome.org/mailman/listinfo/desktop-devel-list

Re: Bugzilla migration tool user accounts

2017-11-30 Thread Alexandre Franke
On Thu, Nov 30, 2017 at 6:43 AM,   wrote:
> Bastien, do you also mean to imply that the wish to be subscribed to >2000
> issues means that receiving >2000 mails won't be a problem?

It will be a problem, but not as bad as losing subscription on those
issues. It would still be very much worth looking for a way to
temporarily disable notifications to avoid that flood. I can also
imagine that sending that many email out at once can raise some flags
and make GNOME look bad in the eye of some email providers.

-- 
Alexandre Franke
GNOME Hacker & Foundation Director
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
https://mail.gnome.org/mailman/listinfo/desktop-devel-list


Re: Bugzilla migration tool user accounts

2017-11-30 Thread Bastien Nocera
On Thu, 2017-11-30 at 05:43 +, philip.chime...@gmail.com wrote:
> On Tue, Nov 28, 2017 at 6:48 AM Nicolas Dufresne  a> wrote:
> > Le mardi 28 novembre 2017 à 14:56 +0100, Bastien Nocera a écrit :
> > > Hey,
> > >
> > > On Tue, 2017-11-28 at 04:21 +, philip.chime...@gmail.com
> > wrote:
> > > > Hi list,
> > > >
> > > > This is about the migration tool for Bugzilla bug reports to
> > GitLab
> > > > issues. See, for background, issues [1] and [2]. Currently, the
> > > > migration tool posts all the migrated issues and their
> > associated
> > > > comments as one single user; either the maintainer who runs the
> > > > script, or a special 'bugzilla-migration' user.
> > >
> > > At the very least, I'd expect people who were on CC: for the
> > Bugzilla
> > > bugs to also be CC:ed on the GitLab issues.
> > >
> > > I'm currently on the CC: list for more than 2000 bugs in
> > Bugzilla. That
> > > doesn't cover modules for which I'm a co-maintainer and would
> > have
> > > received automated -ma...@gnome.bugs emails. I would hate to have
> > to
> > > resubscribe to every one of those by hand.
> > >
> > > I already found the 20-odd bugs I was CC:ed on in gnome-calendar
> > a pain
> > > to resubscribe to, so you can imagine with 2k bugs.
> > >
> > > Also, my GNOME.org gitlab registered e-mail address isn't the
> > same as
> > > my bugzilla mail address. It's another problem you might run into
> > with
> > > the migration script.
> > 
> > To be checked in the script, but you have to make sure your
> > bugzilla
> > email is listed in gitlab (it supports multiple email). After
> > migration
> > there will be no issue removing old emails. You can also change
> > your
> > email in bugzilla too if you want, that works these days.
> 
> Indeed, I'm not planning to support an outside mapping of emails
> other than what is in GitLab — good catch that GitLab accounts can
> have more than one email address. I'll need to check that the script
> will behave properly in that case.
> 
> Bastien, do you also mean to imply that the wish to be subscribed to
> >2000 issues means that receiving >2000 mails won't be a problem?

That's fine if I only receive a single mail for each issue. It's either
that, or having to resub to each issue, or lose track of issues I'm
interested in, neither of which are particularly appealing.
___
desktop-devel-list mailing list
desktop-devel-list@gnome.org
https://mail.gnome.org/mailman/listinfo/desktop-devel-list