Mailing List Archive

Weird screw-up of diffs
Have a look at the history of [[Pim Fortuyn]]

http://www.wikipedia.com/wiki/Pim_Fortuyn&action=history

All versions are properly listed and stored correctly as far as I can tell.
However, starting with the version of Tuesday May 7, 8:59, the diffs
are screwed up. They do not compare to the next version, but to very
old versions.

I noticed that on the Talk page, Tim Shell explained one of his
edits to the main article early Tuesday morning, and I remember seeing
that edit (taking out the label "right wing"). This change however
does not show up on the main page's history page.

Does anybody have an idea what is going on here? Are similar things
happening on other pages?

Axel
Re: Weird screw-up of diffs [ In reply to ]
On sab, 2002-05-11 at 15:03, Axel Boldt wrote:
> Have a look at the history of [[Pim Fortuyn]]
>
> http://www.wikipedia.com/wiki/Pim_Fortuyn&action=history
>
> All versions are properly listed and stored correctly as far as I can tell.
> However, starting with the version of Tuesday May 7, 8:59, the diffs
> are screwed up. They do not compare to the next version, but to very
> old versions.
>
> I noticed that on the Talk page, Tim Shell explained one of his
> edits to the main article early Tuesday morning, and I remember seeing
> that edit (taking out the label "right wing"). This change however
> does not show up on the main page's history page.
>
> Does anybody have an idea what is going on here?

Short answer: there seems to be a break in the linked list of stored
edits.

The history list mostly ignores the linking and just orders by
timestamp, but when doing a diff on the most recent version, the
cur_old_version field is used to select which entry from the old table
to compare from. If that field is pointing to the wrong place, it'll
show the wrong version.

(The $64,000 question is: why is it wrong? I don't know offhand, but I'd
check the code for saving an article. As I recall, the bit that checks
which old_version ID to insert seemed a bit fragile last I looked at
it.)

Here're the item IDs for Pim_Fortuyn; old_id is a unique identifier, and
(cur|old)_old_version is supposed to reference the id of the previous
edit:

old_timestamp old_id old_old_version
20020324153744 36848 0
20020324160320 37456 36848
20020325143818 38206 37456
20020326133630 38299 38206
20020326153001 38302 38299
20020326154453 38305 38302
20020326154558 38307 38305
20020326155234 48454 38307
20020408152422 63034 48454
20020506110024 63063 63034
20020506130808 63123 63063
20020506162701 63126 63123
20020506164202 63129 63126
20020506164649 63130 63129
20020506165926 63284 63130 ^- Good
20020507065958 63367 36848 v- Bad
20020507104949 63372 37456
20020507114506 63426 38206
20020507154732 63428 38299
20020507155020 63429 38302
20020507155211 63431 38305
20020507161527 63453 38307
20020507174244 63519 48454
20020507201257 63775 63034
20020508092908 64614 63063
20020510210808 64721 63123
20020511045521 64847 63126
20020511140710 64855 63129

cur_timestamp cur_old_version
20020511143100 63130

Things seem to have gone awry in the case of this article the morning of
May 7.

> Are similar things
> happening on other pages?

I wouldn't doubt it.

-- brion vibber (brion @ pobox.com)
Re: Weird screw-up of diffs [ In reply to ]
Even with the code in perfect shape (yeah, sure;) I doubt the system
(especially the database) would stay in flawless order for an unlimited time.
So, as suggested before, there should be a "maintnance" script that takes
care of fixing bugs that are known to occur (think of the linked/unlinked
entries). It could be run once a week, or on special invocation of a sysop,
if needed.

Magnus

Am Sonntag, 12. Mai 2002 01:57 schrieben Sie:
> On sab, 2002-05-11 at 15:03, Axel Boldt wrote:
> > Have a look at the history of [[Pim Fortuyn]]
> >
> > http://www.wikipedia.com/wiki/Pim_Fortuyn&action=history
> >
> > All versions are properly listed and stored correctly as far as I can
> > tell. However, starting with the version of Tuesday May 7, 8:59, the
> > diffs are screwed up. They do not compare to the next version, but to
> > very old versions.
> >
> > I noticed that on the Talk page, Tim Shell explained one of his
> > edits to the main article early Tuesday morning, and I remember seeing
> > that edit (taking out the label "right wing"). This change however
> > does not show up on the main page's history page.
> >
> > Does anybody have an idea what is going on here?
>
> Short answer: there seems to be a break in the linked list of stored
> edits.
>
> The history list mostly ignores the linking and just orders by
> timestamp, but when doing a diff on the most recent version, the
> cur_old_version field is used to select which entry from the old table
> to compare from. If that field is pointing to the wrong place, it'll
> show the wrong version.
>
> (The $64,000 question is: why is it wrong? I don't know offhand, but I'd
> check the code for saving an article. As I recall, the bit that checks
> which old_version ID to insert seemed a bit fragile last I looked at
> it.)
>
> Here're the item IDs for Pim_Fortuyn; old_id is a unique identifier, and
> (cur|old)_old_version is supposed to reference the id of the previous
> edit:
>
> old_timestamp old_id old_old_version
> 20020324153744 36848 0
> 20020324160320 37456 36848
> 20020325143818 38206 37456
> 20020326133630 38299 38206
> 20020326153001 38302 38299
> 20020326154453 38305 38302
> 20020326154558 38307 38305
> 20020326155234 48454 38307
> 20020408152422 63034 48454
> 20020506110024 63063 63034
> 20020506130808 63123 63063
> 20020506162701 63126 63123
> 20020506164202 63129 63126
> 20020506164649 63130 63129
> 20020506165926 63284 63130 ^- Good
> 20020507065958 63367 36848 v- Bad
> 20020507104949 63372 37456
> 20020507114506 63426 38206
> 20020507154732 63428 38299
> 20020507155020 63429 38302
> 20020507155211 63431 38305
> 20020507161527 63453 38307
> 20020507174244 63519 48454
> 20020507201257 63775 63034
> 20020508092908 64614 63063
> 20020510210808 64721 63123
> 20020511045521 64847 63126
> 20020511140710 64855 63129
>
> cur_timestamp cur_old_version
> 20020511143100 63130
>
> Things seem to have gone awry in the case of this article the morning of
> May 7.
>
> > Are similar things
> > happening on other pages?
>
> I wouldn't doubt it.
>
> -- brion vibber (brion @ pobox.com)
>
> _______________________________________________
> Wikitech-l mailing list
> Wikitech-l@ross.bomis.com
> http://ross.bomis.com/mailman/listinfo/wikitech-l
Re: Weird screw-up of diffs [ In reply to ]
I will quickly and urgently run a command line script to recreate the missing/broken
tables, if someone sends it to me.

Wouldn't it be essentially the same as the script I ran to create those tables in the
first place?

--Jimbo