Page MenuHomePhabricator

Bizarre page blankings
Closed, DeclinedPublic

Description

Author: timwi

Description:
BUG MIGRATED FROM SOURCEFORGE
http://sourceforge.net/tracker/index.php?func=detail&aid=930934&group_id=34373&atid=411192
Originally submitted by Nobody/Anonymous - nobody 2004-04-07 10:52

Unfortunately, I have not been able to consistently
reproduce this yet. (It seems to only occur when an
error in submitting is involved, and may also involve
navigational buttons.) I am using MSIE 6 on Windows
XP, and basically default Wikipedia user settings.

I have only seen this problem occur twice so far, to my
knowledge. One time was while editing the article
http://en.wikipedia.org/wiki/Dead_man's_switch ; I was
trying to correct a typo, which I marked as a minor edit.
However, the system rejected it, citing an editing
conflict. I looked at the history and saw no conflict; the
last post was made quite some time ago. I forget the
exact events that followed, but I remember I hit the
back button on my browser, made my correction again,
and submitted it, and came across another editing
conflict; this one was genuine. It turns out I had
somehow blanked the page, and somebody already
managed to catch it and revert it (hence the second
conflict, as at the same time I was trying to get things
in control). Note that I still saw no conflict in the history
at the time the first one supposedly happened; only one
other person had posted an edit. Thus, that problem
was not simply that my browser wasn't showing any
edits since I had last seen the page.

The history shows something like this: I made my typo
correction (NOT marked as a minor edit, but I specifically
told it to make it one), then I "blanked" the page
(marked as a minor edit). Note that the blanking has the
description "typo", which, like the attribution of a minor
edit, was supposed to be part of the correction, not the
blanking, which shouldn't be there at all.

The second incident is as follows; I didn't realize this
happened at all because it corrected itself. I was making
various edits to the article
http://en.wikipedia.org/wiki/Prettyprint . The last three
in the history show me fixing a typo I made (marked as a
minor edit, and with the description "fixed typo I made").
Then the next one shows me blanking the page, which
again I had no intention of doing, and this is marked as a
non-minor edit. Then the last "edit" I "made" reverts to
two edits ago, complete with the same description
of "fixed typo I made" and marking as a minor edit. I
seem to recall having gotten an error here, too, so I
SUSPECT what happened in this case was this:

  1. Tried to correct a typo. Marked as a minor edit and

typed in description.

  1. Clicked submit.
  2. Got an error message; hit back button on browser.

The page is blanked without my knowledge <i>after</i>
having been updated successfully.

  1. The text boxes and check boxes still contain my

changes, so I press submit again. Thus, my edit goes
through and appears identical to the one I originally
posted.

Sorry this is has been so long-winded, but I'm really tired
and am unable to condense my text well. Yet, I feel I
must report this now lest I forget more details. Sorry for
the inconvenience. :)

  • Additional comments ------------------------

Date: 2004-04-07 10:58
Sender: SF user furrykef

Sorry, I was the poster of this item. I did try to log in before
writing my post (and thought I succeeded), but somehow I
wasn't logged in when I actually submitted. SourceForge is
really quirky for me that way...it's hard for me to stay logged

in for some reason. Oh, well. :)

Date: 2004-04-14 12:39
Sender: SF user furrykef

I had this problem again, this time with wikibooks. This almost
certainly has nothing to do with navigational buttons (I didn't
use them this time) and everything to do with getting an error
upon submitting. I don't know why my submission goes
through and it raises an error anyway, though. The problem
MIGHT be in MSIE, and/or involve my unreliable
connection...unfortunately, I don't know how to replicate

conditions that produce this kind of error.

Date: 2004-04-14 17:39
Sender: SF user hashar

Just got this trouble on fr.wikipedia.org:

The blank can be seen at:
http://fr.wikipedia.org/w/wiki.phtml?title=Si%C3%A8ge_d%
27Al%C3%A9sia&diff=0&oldid=372643

I was just previewing a modification (seems to have triggered
a page save) then saved it with comment "fix". Checking
history it seems the preview have been save somehow.


Version: unspecified
Severity: normal

Details

Reference
bz317

Event Timeline

bzimport raised the priority of this task from to Medium.Nov 21 2014, 6:47 PM
bzimport set Reference to bz317.
bzimport added a subscriber: Unknown Object (MLST).

rowan.collins wrote:

I strongly suspect this of being another face of bug 275: although that one
describes duplication, and this one blanking, both seem to hinge on some kind of
edit conflict with yourself. In the case of this one, the conflict is clearly
inadvertent - due to some quirk of network or data flow, or just accidental
double-clicking, I'm not sure. I have a feeling this bug may be very hard to
replicate.

timo wrote:

Suggestion: Close the bug ;)

It's open for over one year now - probably a temporary outage/problem or
something that has been fixed already.

Matt_Domsch wrote:

I ran into exactly this behavior twice in the past week, once with a 1.3.x
install, and now with a 1.5.x install. For me, it was clicking "Save Page",
then "back", then "Save Page" again. After this, referencing with only the page
name would cause a blank page to be shown. Referencing with the page name plus
some text (i.e. MyPage#foo) would cause the page to be shown.

Not a bug, Matt, you are trying to view the blank page from your local browser's cache!
the sequence "Save", "back", "Save" will perform the first save for your self, then when clicking back you
are viewing the edit page before your change but with the same edit id, and the second save detects the
conflict with yourself.

There's no way a Save/back/save sequence can work without generating a normal edit conflict: only one save
is allowed per edit.

I've closed this bug as suggested in comment #2. The last reported sighting of this bug was over three years ago, and the bug comment #1 suspected as the cause has since been resolved.