I'm mainly talking about higher profile pages. I looked up George Washington. It has
around 50 templates and 17 images. That would require MySQL to scan up to 134 rows on each
page view to see if things where changed, and always 134 if you wanted how many
templates/images where changed. That's not really worth it.
The UI change was not much better, and it even through of Phillip resulting in some
confusion earlier.
There are performance constraints, but also this issue is very much semantic, and I've
been aware of it for months.
Before adding band aids or other quick changes, some things need to be though out:
1) What will most people think "revision" and "change" means?
Probably they would thing of either as any change to the content. The MW jargon for
revision means wikitext of the text corresponding to a page revision. This ignores
templates/images and readers will not get it even if we try to explain.
2) Is that the word we want to use to describe changes?
Since revision is already confusing jargon for newbies, I'd prefer
"changes". Any attempt to use any word that means the same as the MW jargon word
"revision" will confuse people.
3) So we should always be talking about complete visual states of articles rather than the
jargon of "changes not counting templates/images"?
Ideally, that would be nice.
4) How can we refer to any change (including templates/images) without causing unneeded
performance drops?
This is the hard question. Currently, the UIs say "X changes need review" or
the like. Changes means "MW revision jargon". This is bad. If we *want* to keep
stating the number, and refer to ALL changes (templates/images too), that is not
efficient, though it is technically possible. You would have to find each image/template
that has a newer version the the id/sha_1 assigned to the stable version of the page in
question and the run queries on each of those to count how many changes have been made.
For George Washington, well, it's a nightmare ;). If want to keep the number, then it
most only refer the jargon "revision" word ...
5) Do we want people re-reviewing every time a template is updated?
This would be encouraged if we declared on the tags whether templates are out of sync
or not and we wanted to always keep it 100% up to date. I'd actually be confident that
most articles get enough edits (which updates templates when autoreviewed or prompts the
user to review) that they would stay reasonably up to date. My concern is that if people
are constantly re-reviewing a page revision to update templates/images, that the fr_text
column will bloat up too much (it will be sizable no matter what, but that would make it
worse).
So I'd like to treat uses of the word "change" as what readers would expect:
any change to the page. However, I have yet to find a good way to do so.
-Aaron Schulz
Date: Sun, 7 Oct 2007 03:52:25 +0200
From: erik(a)wikimedia.org
To: wikiquality-l(a)lists.wikimedia.org
Subject: Re: [Wikiquality-l] Reverts
On 10/6/07, Aaron Schulz <jschulz_4587(a)msn.com> wrote:
In that case, even one change to a template
would make it not show. I fear
that on wikipedia, that condition would rarely show up, and I'm not sure
it's worth the performance hit.
I just did a quick sample of en.wp articles. Most have a template of
some kind, but most high-use templates are semi-protected, which means
they're only likely to be edited by users in the editor group anyway.
Those which aren't are typically not edited much at all. So how can
you conclude that the "sighted and current" condition would rarely
show up? Indeed, everything leads me to conclude that it's the _most
common_ condition; it's the one that we want to achieve on everything
through ever-improving patrolling mechanism.
To then not show a clear visual indication of the sighted status
utterly confounds me from a usability perspective. You end up with an
inconsistent UI where icons sometimes show up and sometimes don't with
no clear reason why. And you want to explain this with: "Oh, BTW, if
you look at the number of changes, and it's 0, then actually, not
taking templates into account, the main body of the article has been
sighted"? That's a usability nightmare. I found this issue so major
that I wasn't even comfortable with taking the extension demo live
before at least applying some band-aid.
I do agree that we should make sure it works correctly depending on
the template status. Which performance hit are you worried about
specifically? Shouldn't it just be a matter of looking up the flagging
status of the templatelinks associated with the page in question?
--
Toward Peace, Love & Progress:
Erik
DISCLAIMER: This message does not represent an official position of
the Wikimedia Foundation or its Board of Trustees.
_______________________________________________
Wikiquality-l mailing list
Wikiquality-l(a)lists.wikimedia.org
http://lists.wikimedia.org/mailman/listinfo/wikiquality-l
_________________________________________________________________
Boo! Scare away worms, viruses and so much more! Try Windows Live OneCare!
http://onecare.live.com/standard/en-us/purchase/trial.aspx?s_cid=wl_hotmail…