Jump to content

Wikipedia:Squashed bugs

From Wikipedia, the free encyclopedia
This is an old revision of this page, as edited by AxelBoldt (talk | contribs) at 01:05, 26 February 2002. The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

Please report any new bugs in the Wikipedia software over in wikipedia:Bug Reports. This is not the place to point out mistakes in a Wikipedia article (you can do that by simply editing that article), nor is it the place to request new software features (you can do that at Feature requests), nor is it the place to report new bugs, because nobody's going to notice them here.

Only put a bug here if it's been resolved, and you're taking it out of the Bug Reports page!

Fixed


Edit conflicts with myself

I seem to be getting into regular edit conflicts with myself! I start editing a page and it throws up the edit conflict dialog, but top and bottom boxes are identical and contain only my immortal prose. a second "save" usually fixes this, except, it seems, when I am working on a redirect, when I end up in an infinite loop of edit conflicts. user:clasqm

This happened to me too. It doesn't seem to happen as much anymore, though...maybe it has been fixed? Dreamyshade (2002/1/31) seems as it has been fixed -- ChaTo

No, it has not been fixed. See Mercury/God, for example. --Zundark, 2002 Feb 3
I don't have any problem editing that page now. Can you confirm? --Brion VIBBER 2002-2-6 23:09 PST
This is fixed now, but it was still broken on Feb 3. --Zundark, 2002 Feb 22

RFC auto-linking no longer works (2002/02/09)

I seem to recall that writing, say, RFC 1519 in an article would auto-link to a repository of RFC's. This no longer seems to work.

IMHO, that's a feature, not a bug. --Robert Merkel
However, the loss of a feature that once existed is a regression on a feature request. Are those bugs? --Damian Yerrick
Yes. --The Cunctator
I'm arguing that the riginal feature is inappropriate for Wikipedia and should not be reinstated. --Robert Merkel

Parser does not escape HTML

Also, some HTML in page comments is not being escaped out - for example, put <p> in a page comment, and see what happens. -- The Anome

Should be fixed in the next upgrade... --Brion VIBBER 2002/02/06 01:29 PST

The parser treats white space as significant, which is undesirable. For example, "=== Header ===" will normally produce a level three header line, but if it is followed by whitespace, it will produce a level two header line surrounded by equal signs.

Furthermore, the new parser treats equal signs as a header line even if they occur in the middle of some other text. They should only be treated as a header line if they occur on a line all by themselves. AxelBoldt

The Formula One page has some problems with the section headings (with the equals sign) -- Jheijmans

There are similar problems at Surreal numbers when the string "<=" is used. -- Jan Hidders

There are also issues with the processing of '==' in this example. See also Fermats little theorem for another example of layout damage caused by over-aggressive matching of pairs of '=' characters to generate markup. -- The Anome

I dont think the '====' surrounding the first word of an article works right. Maybe nobody uses this any more? It was in the how to edit a page page. Also, doing the '===' surrounding subsections works some of the time. shhhh don't hack my ip! BF *)

    • (3 Febr 2002) A patch has been submitted for this, so this should all work after the next update of the software. -- Jan Hidders~

Lower case user names


Not sure how many other people are affected by this, but my "Contributions" page is totally empty. Since the changeover I have edited over 30 articles and started several new ones. You can see what I mean at http://www.wikipedia.com/wiki.phtml?title=special:contributions&amp;amp;theuser=Maveric149. --maveric149

BTW, http://www.wikipedia.com/wiki.phtml?title=special:contributions&amp;amp;theuser=maveric149 this works. --Chuck Smith

Yep, second link does work. The only difference is that my user name is lower case in Chucks example. I created my account with a lowercase "m", and that is how I log in - yet when I click on "This user's contributions" in my default userpage, I get zip. Thanks for the info Chuck! --maveric149

Fix is ready, will be installed soon. --Magnus Manske

(2002/1/1) The first letter of usernames are automatically capitalized upon submission of the "preferences" page.

If a all lowercase string is input into the "UserName" field in the preferences page, the first letter of said string is capitalized.

All page titles in Wikipedia should be capitalized. Suggest resolving INVALID. --Damian Yerrick

I have a similar problem - there's a user jheijmans (small j) and Jheijmans (capital J); both user pages link to the same page, but the contributions page does not - can this be merged into one user entry? -- Jheijmans

They are the same page; see contribution list bug above. --Magnus Manske

'New separators: Horrible'

(2002/2/7) Just noticed a black line separating page's parts. I think it's horrible, i prefer without that. Maybe we need a option for that. -- Tuxisuau


Special Pages

Most Wanted Pages not current (2002/02/10)

A third of the articles listed on special:WantedPages already exist. Clicking on the link to regenerate the page doesn't make them go away.

(2002-02-12) By now, over 70 percent of the Most Wanted pages have been filled. When I tried to refresh the page to bring in more entries, I got the following error:

(The page was last refreshed just -160980 minutes ago; please wait another 160985 minutes and try again.)

160985 minutes equals 111 days and change. Was this put in place on purpose, to keep people from refreshing the page until early June? After a while, the feature becomes absolutely useless. --Damian Yerrick

The timestamp is incorrect and needs to be fixed by someone with access to the database. --Brion VIBBER
In the meantime, it would be better to remove the "Mosted wanted" page altogether, as all the articles it lists now exist. --Zundark, 2002 Feb 20

"Most Wanted"

There appear to be a lot of articles listed in "Most wanted" which actually do exist. For example, Year in Review/Guidelines, 1, 2, 3, etc.

Just to follow up this one with a further observation, it looks like the Year in Review/Guidelines link listed in Most wanted is for Year_in_Review/Guidelines_, with a space at the end after "Guidelines." Shouldn't trailing spaces be trimmed from page titles and links?

(2002/1/26) The "Most wanted" page indicates that a whole bunch of pages want a "Year in Review/Guidelines" page, but that page already exists. Upon poking around a bit more, I found that all of the pages who supposedly had a link to the non-existant page had links with a trailing space, and this was being considered different from the link without a trailing space. For example:

(2002/1) The "wanted" page considers the second link to point to a non-existant page. At time of writing, the "most wanted" page contains a lot of Year in review pages that do actually exist. example: 12 and 17. clasqm

indeed, the 'most wanted' page is now swamped with two-digit year in review entries. Somethings up, since they do exist and are linked. --MichaelTinkler


HTML named entities not working while previewing (2002/02/09)

HTML named entities are escaped while previewing and because of that they don't work (e.g. &amp;amp;deg; becomes &amp;amp; d e g ;). See Jupiter for an example. --Uriyan

They're only escaped in the preview render; in the article text they remain correct, and will be shown properly after saving. Should be fixed soon... Brion VIBBER


: in page titles (2002/02/08)

In putting titles of plays into the Pulitzer Prize page, some of the letters which are capitalized on the editing page appear in the final text as being uncapitalized. The example I noticed was for Angels in America: Millennium Approaches, in which the first two capitalized A's are not capitalized in final text even though they are capitalized on the editing page (as is true on this page as well}.

Sorry, you can't use colons (:) in page titles. The colon is reserved for special pages, like special:Watchlist, talk:France, log:Uploads, etc. "angels in america" is getting put into lowercase because that's the way special pages are named, with the first part in lowercase. I recommend using Angels in America - Millennium Approaches instead to avoid this problem. (See the PHP script FAQ for more about these niggling little things.) --Brion VIBBER

Recent changes shows previous edit (2002/02/09)

Edit a page, and recent changes shows it as the previous edit; I noticed this when I'd apparently just edited a page I hadn't touched for a week; someone else had just edited it, buit it appeared in Recent changes as my previous edit, complet with summary.


Just submitted this edit, and the edit shows OK. Looks as if the sequence is actually: someone creates a new article; No problem. Someone comes along, edits the new article, which has had no revisions yet, and then recent changes reports that edit as the original creation of the article instead...

Sounds like someone created a new page that your page linked to. Known bug, fixed in CVS, fix should show up here soon. --Brion VIBBER

Unable to edit large pages )2002/02/09)

when I open to edit any of the larger pages (eg Nuclear Weapon, Wikipedia: Bug reports) I am unable to type any characters in. I can move the cursor and cut-paste but I cannot type.

What browser, what version, what operating system are you using? (If you're using Mozilla, try going into your preferences and disabling the QuickBar. There's a bug in Mozilla that makes editing a text box inside a table verrrrrrry slowwwwww.) Brion VIBBER

Timestamps for first two previous version links in a page's History are always the current time

(2002/1/27) Check the History link for this page... -- user:Derek Ross

Timestamps for previous version links in a page's History are off by one and the first two are therefore set to the current time.

Check the History link for this page... -- user:Derek Ross

Bad search results This probably has been said already somewhere by somebody, but I'll put it here. If I search for example for 'Rome', I get a lot of results, begining with AIDS (syndROME), Alfa ROMEo and AndROMEda. while Rome, the city, would be an exact match. Could it be made that exact search results are listed first, or maybe there can an approximation measure for the matches? user:Jheijmans

The search engine does not generate any results using Netscape 3.0 and junkbuster. GWO


Wiki.phtml pages

I accidently created a wiki.phtml page with IP address 130.94.122.xxx after hitting a []? link. there are a number of other examples of this: September 11, 2001 Terrorist Attack/wiki.phtml; English language/wiki.phtml;

Complete list of encyclopedia topics/wiki.phtml.


Performance

Love the new functionality, but I haven't been able to contribute nearly as much as I would like because the pages come up at a sub-glacial speed (very often). Have broadband &amp;amp; fast computers at home, work and school and I have the same problem (every other webpage I go to zips and zings --- including heavily hit and often

Appears mostly fixed as of 2002/02/20. It's now much faster than Everything2. --Damian Yerrick


Character 0 vanishing (2002/02/07)

The second line of http://wikipedia.com/wiki/Geologic_timescale starts off with " 0 ". The 0 is simply disappearing in the displayed output. I put in an extra space to make the page look closer to right. I'm reasonably sure the 0 was being displayed two weeks ago.

Fixed in CVS; should work again when the software is next updated. --Brion VIBBER 2002-02-07

Bogus Logged IP (2002/02/09)

It's really easy to log a fake IP address, see [1] for example. I'm not going to just explain how here, though; anyone who knows how the previous logged IP address problem was fixed should be able to figure it out, or can email me. --Carey Evans

You don't even need to fake it, Wikipeida is logging me as 213.105.195.101, while the reality is 62.62.253.xxx.

It seems to be fixed now; presumably it now takes the correct host from X-Forwarded-For if more than one is there. --Carey

Full IP address in page history (2002/02/09)

Although the last octet of the user's IP address is xxx'ed out on Recent Changes, the page history now shows the complete IP address. --Carey Evans

Bug, or feature? --Damian Yerrick
It's changed back now, so I guess it was a bug. --Carey

(2/13/2002) The Recent Changes page is much better fast, but it's fast because it's cached. It should be fully updated with the last edits made; the current lag is pretty extreme. --The Cunctator

I'm not seeing this lag. More detail please? --Brion VIBBER



Login and Preferences

Edits when not logged in are considered "minor"? This, in my mind, is a horrible idea. I hope that's a bug, not a feature. Koyaanis Qatsi

Ack! That's the exact opposite of what it's supposed to do. Expect this to be fixed sometime soon... Brion VIBBER 2002/2/10


Cookie problems


Hey, Jimbo -- I log in, hit 'preferences' and it tells me I am not logged in. So...I'm not going to preferences for a while! J Hofmann Kemp

I have this same problem, no matter what I hit after I log in. It doesn't seem to recognize my login. Cookies are enabled on my browser. RjLesch.

follow-up to above: my login worked fine when I selected the "remember my password as a cookie" option. User:RjLesch
follow-up to above: my login worked fine when I selected the "remember my password as a cookie" option. User:RjLesch
I also run Moz 0.9.7 on Debian unstable, no problems for me. --User:David Merrill

Login problems: I can successfully log in, but as soon as I go to another page (for instance, the preference page), the system thinks I'm not logged in. --Robert Merkel (whose ip address starts with 144.132.75)

I get this behavior if cookies are disabled in the browser; the script prints out a "logged in successfully" message based on your username/password matching, ignoring the setting of the cookie, so the failure is silent. --Brion Vibber
My browser (Moz 0.9.7, part of Debian unstable) is set to ask for approval to store a cookie. No dialog indicating the receipt of a cookie pops up, and there's no cookies from wikipedia.com in my saved cookies. --Robert Merkel
I tried it on several different machines, at home and at work, using IE5.5, 6.0, Netscape (Windows), Netscape (Linux), Opera (Windows), and Konqueror, and they all work just fine with the cookies. --Magnus Manske

Here's the scenario. I had a password for my account saved to cookie with the checkbox on the log in page. I then changed my password to the old admin password (so now you techies know what my password is) and now when I try to login (again with cookies on) it says that I successfully login, but when I try to do anything, it shows that I'm not logged in! HELP!!! --User:Chuck Smith

Feb 2, 2002: On Linux/Debian with Netscape 4.76, after logging in, it forgets that I'm logged in immediately after I change to another page from the login page. This happens whether I check "Remember my password with a cookie" or not. Maybe it is related to the next bug below? AxelBoldt


Nested tables don't work (2002/02/23)

While trying to come up with a template for the table to be used in all chemical element entries, I discovered that the wikipedia software appears to deliberately break table code that should work just fine. I was attempting to create a table that had 2 columns in the top half and 6 columns in the bottom half, with the 6 column half handling its internal layout independantly of the 2 column half; to this end, I created a 6-column table and nested it inside the 2-column table. It worked just fine when I tested it outside of Wikipedia. However, upon testing it in wikipedia, I got the following:


Properties
General
NameBeryllium
SymbolBe
SeriesAlkaline Earth Metal
Group, Period, Block2 (IIA), 2, s
3rd ionization potential14848.7 kJ/mol (153.893 eV)
Isotopic
isoNAhalf-life (t1/2)DMDE (MeV)DP
7Be{syn.}53.12 daysepsilon0.8627Li
9Be100%Be is stable with 4 neutrons
10Betrace1.51 &amp;times;106 ybeta-0.556B10

It appears that the second set of &amp;lt;table&amp;gt; tags got escaped.


Wrong links with cross-language links (2002/02/05)

On pages with cross language links, like Table tennis, all the links on the page show as external links, even though they're to Wikipedia articles. --Carey Evans

Fixed in CVS, should be up sometime soon along with some HTML fixes and improvements to the search engine. 2002/2/14 Brion VIBBER


Javascript in generated pages

(2002/1/27) Too much HTML is let through unchallenged. For example, try the following link not to my user page, Carey Evans. HTML 2.0 plus tables should be enough for formatting Wiki pages, though I'd prefer to be able to do everything without HTML at all. --Carey Evans

This is partially fixed; non-approved HTML tags are no longer let through, but they're not checked for internal safety. Ie, I can still do the same thing using a "safe" tag. JavaScript-releated elements should definitely be removed... --Brion VIBBER 2002/02/06

Here's another example of evil javascript I found over in the Goatsec.xs article. It used to lead to the web page in question, but I've rewritten it to spare your eyeballs. Just move your mouse pointer over the link and then without clicking it move your mose pointer off of it again. http://www.wikipedia.com/wiki/Goatse.cx


HTML tags should not be possible in the page title

(2002/1/29) See http://www.wikipedia.com/wiki.phtml?title=Christopher_Tolkien for example. This shows up in the recent changes, where the html is rendered if the page is edited. Can do some nasty stuff with unclosed tags. This is in konqueror 2.2, maybe the effect is somewhat different in other browsers.

Nope, it's a bug. Working on it. --Magnus Manske



Preview of &amp;lt;pre&amp;gt; and handling of HTML entities

(2002/1/29) The preview of a page which has HTML entities inside a &amp;lt;pre&amp;gt; environment differs from the rendering of the saved page. Specifically, if the page contains the text &amp;amp; a m p ; (without the spaces) inside a &amp;lt;pre&amp;gt; environment, then the preview will show this as &amp;amp; a m p ; while the saved page will render it as &amp;amp;. This is an issue when editing the "Character Formatting" section of How to edit a page. --AxelBoldt


ISBN numbers with check digit of X (2002/02/10)

ISBN numbers with a check digit of "X" like ISBN 1-55634-399-X are not linked correctly to Pricescan - the numeric part is linked, but a space is inserted before the "X" and it is not included. --Carey Evans



Edit Conflict! with myself! and creation of new, incorrectly titled pages (2002/02/10)

When I try to redirect, I get an edit conflict BUT, just to make it interesting, I'm also somehow creating a new page. Here's what happens:

  1. I'm logged in
  2. I open a page (this seems to be happening only when I'm opening a page on a titled person, BTW) to edit
  3. I copy the information in the page and then type #REDIRECT pagename
  4. I get an edit conflict

BUT -- a new page has been created, with a funny Bpagename/b rest of pagename as the title.

Does the link to your new page have bold tags (&amp;lt;b&amp;gt;) in it? Knowing *exactly* which pages you're talking about would be a great help. -- Brion VIBBER

sorry Brion -- you mighe want to check the recent changes, but the ones I can think of off the top of my head are Gustavus Adolphus (AKA BGustavus Adolphus/b) and BMaximilian I/b holy Roman Emperor -- I viewed the source, and for some reason there are complete HTML tags in the code -- these appear as I've just described on the page titles. David Parker and I think Lars have re-created the pages (namespaces?) properly by adding the correct title (with underscores in the spaces) after '/' in the navigation bar. JHK

Okay, my guess at what's going on is that someone created the page from a link that turned up in the search page. The search page is horribly buggy and, among other mortal sins, puts bold tags around located search terms *before* the page extracts are wiki-interpreted, thus creating links with bold tags in them. Previously, the tags would actually slip into the page title, sometimes unnoticed. Now, the &amp;lt; and &amp;gt; are stripped out, the B and /B are clearly visible, and it should be noticible to the person editing the page. In the short term, please read the title when you're editing a new page! If it looks completely wrong, start over. The search page is being rewritten, so this can be fixed in the process... Brion VIBBER

This isn't good. Checking now to see if it works if i remove all content first...

Nope -- still happens -- Creates a new page which comes up correctly under search (where the bolded part shows bold), but the tags remain in the actual page title.

I'm not sure I follow. If I go to Bgustav/bus Adolphus of Sweden, I get redirected to Gustavus Adoplus of Sweden and everything looks fine. The messed-up title appears only in the URL bar, because when you follow a redirect you get the new page with the old URL. If I'm looking at the wrong thing, let me know. 2002/2/10 Brion VIBBER


Edit conflict "Text you submitted" not in edit box

(2002/2/12) On the edit conflict page, the the text you submitted should be put in an textarea box, but it's not.

DIFF corrupts the database? (2002/02/08)

The (DIFF) links seem to have several problems. I've noticed that if I click on the first (DIFF) link on either the history page for an article, or on that article's listing on my watchlist, then it refuses to show the difference. When I tried several of the diffs for one article, it actually corrupted the database, I think. As of right now, if you go to Raven_paradox, you'll see what looks like a DIFF page for an old version of the article. But if you click on "edit this page", it brings up the text for the current version of the article. Something must be inconsistent in the database right now.

Known problem, fixed in the code, fix should be online ANY SECOND NOW. Brion VIBBER
It looks like the software isn't damaging pages any more. But the damage to pages like Raven paradox remains. Will that be fixed automatically, or should we start looking for such pages to fix them by hand?
If you see such a broken page, hit "edit this page" then "save", it'll clear the page's cache and all is well. We'll try to put in a way to automatically refresh pages from time to time, but it's not set up yet. Brion VIBBER