Hi All,
Here are the minutes from this week's TechCom meeting:
* TechCom will be announcing a call for nominations to expand the committee next week
* Discussion and community input on MediaWiki Platform Architecture Principles Document continues: https://www.mediawiki.org/wiki/Wikimedia_Technical_Committee/MediaWiki_Platform_Architecture_Principles
* Public IRC Discussion of RFC next week 2018-06-06 in the #wikimedia-office channel at 2pm PST(22:00 UTC, 23:00 CET): Use ar_page_id to determine the parent IDs for undeleted revisions https://phabricator.wikimedia.org/T193690
You can also find our meeting minutes at https://www.mediawiki.org/wiki/Wikimedia_Technical_Committee/Minutes
See also the TechCom RFC board https://phabricator.wikimedia.org/tag/mediawiki-rfcs/.
Am 31.05.2018 um 18:41 schrieb Kate Chapman:
- Public IRC Discussion of RFC next week 2018-06-06 in the
#wikimedia-office channel at 2pm PST(22:00 UTC, 23:00 CET): Use ar_page_id to determine the parent IDs for undeleted revisions https://phabricator.wikimedia.org/T193690
Please note that the scope of the IRC discussion of 2018-06-06 is intended to me more general than what the RFC proposes.
The main questions of the meeting will be:
* do we need rev_parent_id? * if yes, and what exactly should the semantics of rev_parent_id be? * if no, what do we replace it with (if anything)?
I am in the process of preparing a write-up to frame the discussion. I'll post a link here.
-- daniel
As mentioned in the TechCom radar email last week, there will be a public IRC discussion on the future of rev_parent_id on Wednesday, 2018-06-06 in the #wikimedia-office channel at 2pm PST(22:00 UTC, 23:00 CET).
The RFCs original title was "Use ar_page_id to determine the parent IDs for undeleted revisions" https://phabricator.wikimedia.org/T193690. However, the discussion has drifted quite a bit, and is now touching on questions around the size differences shown on the history and contributions pages, which is the only thing that rev_parent_id is used for. Perhaps we should have filed a separate ticket, but at this point, that would just have meant splitting the conversation, and creating more confusion...
Anyway, in preparation of the RFC discussion, here are again the main question that need answering:
* Can we do without rev_parent_id entirely? Can we live without the size info on Special:Contributions, or can we query it cheaply enough?
* Do we want to record the original "edit as intended", and if yes, where?
* On a side note, what's the best way to detect a page's first revision, and do we still need that if we have the page creation log (and recentchanges).
I have tried to summarize the discussion so far, with the available options and considerations, at https://phabricator.wikimedia.org/T193690#4254713.
... the size differences shown on the history and contributions pages,
which is the only thing that rev_parent_id is used for
This may be true in MediaWiki but not so much for external tools. I just wanted to preemptively say this. I'll be joining the IRC discussion to share more :)
~Leon
On Mon, Jun 4, 2018 at 2:23 PM Daniel Kinzler daniel.kinzler@wikimedia.de wrote:
As mentioned in the TechCom radar email last week, there will be a public IRC discussion on the future of rev_parent_id on Wednesday, 2018-06-06 in the #wikimedia-office channel at 2pm PST(22:00 UTC, 23:00 CET).
The RFCs original title was "Use ar_page_id to determine the parent IDs for undeleted revisions" https://phabricator.wikimedia.org/T193690. However, the discussion has drifted quite a bit, and is now touching on questions around the size differences shown on the history and contributions pages, which is the only thing that rev_parent_id is used for. Perhaps we should have filed a separate ticket, but at this point, that would just have meant splitting the conversation, and creating more confusion...
Anyway, in preparation of the RFC discussion, here are again the main question that need answering:
- Can we do without rev_parent_id entirely? Can we live without the size
info on Special:Contributions, or can we query it cheaply enough?
Do we want to record the original "edit as intended", and if yes, where?
On a side note, what's the best way to detect a page's first revision,
and do we still need that if we have the page creation log (and recentchanges).
I have tried to summarize the discussion so far, with the available options and considerations, at https://phabricator.wikimedia.org/T193690#4254713.
-- Daniel Kinzler Principal Platform Engineer
Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Mon, Jun 4, 2018 at 11:54 AM, Leon Ziemba musikanimal@wikimedia.org wrote:
... the size differences shown on the history and contributions pages,
which is the only thing that rev_parent_id is used for
This may be true in MediaWiki but not so much for external tools. I just wanted to preemptively say this. I'll be joining the IRC discussion to share more :)
~Leon
To add to Leon's use case, I would argue that it may be one use case but it's a pretty visible one to active contributors who are familiar with the history page. It may be worth clarifying whether this is a backend change that would keep the feature intact or if this change would result in a change to Wikimedia projects' user experience.
On Mon, Jun 4, 2018 at 2:23 PM Daniel Kinzler <daniel.kinzler@wikimedia.de
wrote:
As mentioned in the TechCom radar email last week, there will be a public IRC discussion on the future of rev_parent_id on Wednesday, 2018-06-06 in
the
#wikimedia-office channel at 2pm PST(22:00 UTC, 23:00 CET).
The RFCs original title was "Use ar_page_id to determine the parent IDs
for
undeleted revisions" https://phabricator.wikimedia.org/T193690. However, the discussion has drifted quite a bit, and is now touching on questions around the size differences shown on the history and contributions pages, which is the only thing that rev_parent_id is used for. Perhaps we should have filed a separate ticket, but at this point, that would just have meant splitting the conversation, and creating more confusion...
Anyway, in preparation of the RFC discussion, here are again the main question that need answering:
- Can we do without rev_parent_id entirely? Can we live without the size
info on Special:Contributions, or can we query it cheaply enough?
- Do we want to record the original "edit as intended", and if yes,
where?
- On a side note, what's the best way to detect a page's first revision,
and do we still need that if we have the page creation log (and recentchanges).
I have tried to summarize the discussion so far, with the available options and considerations, at https://phabricator.wikimedia.org/T193690#4254713.
-- Daniel Kinzler Principal Platform Engineer
Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Am 04.06.2018 um 20:58 schrieb James Hare:
To add to Leon's use case, I would argue that it may be one use case but it's a pretty visible one to active contributors who are familiar with the history page. It may be worth clarifying whether this is a backend change that would keep the feature intact or if this change would result in a change to Wikimedia projects' user experience.
First of all, the purpose of this discussion is mainly to explore the problem and the options we have. We don't expect to approve any action just yet. Perhaps I should have made this clear in my original mail.
The intention is to keep the feature - but since the behavior of the feature is inconsistent (e.g. the size difference shown does not always correspond to the linked diff) and changed over time (due to bugs in the undeletion code, for instance), one important question is what the expected behavior actually is.
Am 04.06.2018 um 20:54 schrieb Leon Ziemba:
... the size differences shown on the history and contributions pages,
which is the only thing that rev_parent_id is used for
This may be true in MediaWiki but not so much for external tools. I just wanted to preemptively say this. I'll be joining the IRC discussion to share more :)
Excellent, thank you! It would be particularly interesting to know what assumptions you make about the semantics of rev_parent_id. E.g. there are three revisions, A, B, and C, and revision B gets romoved - what should revision C's parent be?
Similarly, when revision X gets imported and inserted between A and B, what should revision B's parent be?
On Monday, June 4, 2018, Daniel Kinzler daniel.kinzler@wikimedia.de wrote:
Am 04.06.2018 um 20:54 schrieb Leon Ziemba:
... the size differences shown on the history and contributions pages,
which is the only thing that rev_parent_id is used for
This may be true in MediaWiki but not so much for external tools. I just wanted to preemptively say this. I'll be joining the IRC discussion to share more :)
Excellent, thank you! It would be particularly interesting to know what assumptions you make about the semantics of rev_parent_id. E.g. there are
three
revisions, A, B, and C, and revision B gets romoved - what should
revision C's
parent be?
Similarly, when revision X gets imported and inserted between A and B,
what
should revision B's parent be?
-- Daniel Kinzler Principal Platform Engineer
Wikimedia Deutschland Gesellschaft zur Förderung Freien Wissens e.V.
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
I think in such a case, revision C's parent should benull or 0. Similarly for revision X.
I dont view the parent as what revision came first but what revision was edited to make this revision (i.e. where the current revision was "forked" off from) -- Brian
Excellent, thank you! It would be particularly interesting to know what assumptions you make about the semantics of rev_parent_id. E.g. there
are three
revisions, A, B, and C, and revision B gets romoved - what should
revision C's
parent be?
Similarly, when revision X gets imported and inserted between A and B,
what
should revision B's parent be?
I think in such a case, revision C's parent should benull or 0. Similarly for revision X.
I dont view the parent as what revision came first but what revision was edited to make this revision (i.e. where the current revision was "forked" off from)
Yep, that was my thinking. I was going to save it for the IRC discussion, but since I can say this quickly: I use rev_parent_id to find the literal *previous* revision (revision that was "forked", as Brian says), for the purposes of immediate revert detection and for the diff size. I also do this for the subsequent revision, all with a single query (FROM revision then two JOINs on revision). If I can get this information just as easily by other means, my use case is satisfied.
Also quick side note, I hope it's obvious that viewing diff sizes on revision histories and contribution pages is essential. Indeed it's sometimes wrong, so obviously it'd be great if that could somehow be fixed :) I'm sure you smart people have some great ideas. Looking forward to the IRC discussion.
~Leon
On Mon, Jun 4, 2018 at 3:22 PM, Brian Wolff bawolff@gmail.com wrote:
I dont view the parent as what revision came first but what revision was edited to make this revision (i.e. where the current revision was "forked" off from)
Although that implies that an automatically-resolved edit conflict should have a "parent" as the revision that was edited rather than the current revision, again different from the current behavior (and both of the options Daniel discusses in the task).
It similarly implies that an edit starting with an old revision should record that old revision as the "parent" instead of the current revision, again different from the current behavior. And maybe the same for rollbacks and at least some undos.
wikitech-l@lists.wikimedia.org