Jump to content

User talk:Cyberpower678/Archive 36

Page contents not supported in other languages.
This user uses Twinkle to fight vandalism.
This user has 70% energy left.
This user has administrator privileges on the English Wikipedia.
This user is a global renamer.
This user is an edit filter manager on the English Wikipedia.
This user has interface administrator privileges on the English Wikipedia.
Trout this user
From Wikipedia, the free encyclopedia
Archive 30←Archive 34Archive 35Archive 36Archive 37Archive 38→Archive 40
Senior Editor II
Senior Editor II

IABot errors

In this recent edit, it looks like IABot lost the title and publisher of a citation, and added a meaningless "1=c" parameter. Thanks if you can take a look and debug. —Patrug (talk) 09:14, 17 July 2016 (UTC)

The parsing engine seems to be suffering from a cascading parsing failure on this specific instance. I will need to investigate further.—cyberpowerChat:Limited Access 20:39, 18 July 2016 (UTC)
I've got most of it fixed. Need to fix one more problem before I'm ready to declare it fixed.—cyberpowerChat:Online 16:28, 19 July 2016 (UTC)
 Fixed—cyberpowerChat:Limited Access 19:54, 19 July 2016 (UTC)

Reports

Hey, I'll start making some random checks of IABot and report anything that looks worth bringing to attention. -- GreenC 15:51, 14 July 2016 (UTC)

  • [1] Duplicate URL. I understand how and why this happened tricky problem. Would it be possible to check if the archive URL already exists on the same line and if so not modify the original URL?
    See T140304.—cyberpowerChat:Limited Access 19:38, 14 July 2016 (UTC)
OK I'm even sub'd to it :) -- GreenC 00:26, 15 July 2016 (UTC)
Ah never mind I see CB likely got the archive link from the CB database originating from the wiki database giving it priority over using the IA API which is the right thing. -- GreenC 00:26, 15 July 2016 (UTC)
It's now rescuing the sources when it shouldn't 12 3 -- GreenC 12:55, 18 July 2016 (UTC)
Explain?—cyberpowerChat:Online 15:29, 18 July 2016 (UTC)
In #1, http://en.wikipedia.org/wiki/Wikipedia:Footnotes is a working URL but CB replaced it with an archive URL. -- GreenC 16:34, 18 July 2016 (UTC)
That's because the DB has a `live_state` of 0 for that URL, which tells the bot that it is dead.—cyberpowerChat:Online 17:38, 18 July 2016 (UTC)
Can they be set undead? Not sure which are effected as I only came across these in a spot check. Another option is to blacklist http[s]://en.wikipedia.org/wiki/Wikipedia:* if such a feature exists (like with example.com?). The reason is it is unlikely such a URL would actually be dead and not a redirect or rename or something; and it seems outside the scope of the bot to archive internal system URLs. -- GreenC 21:24, 18 July 2016 (UTC)
Yes. I reset that link already. It's likely a lingering remnant from the trials when the checkIfDead was still being tested and developed. The bot's scope is to handle all external links. If someone is adding Wikipedia links as external links, the bot will check them and MW software is fully compliant with proper HTTP code protocols so there's nothing to worry about. As a side not, IABot is now able to ignore preformatted text.—cyberpowerChat:Limited Access 19:59, 19 July 2016 (UTC)

Shutting bot down. It's editing very fast 50-100 articles every 60 seconds and the above were found in just a 2 minute timeframe meaning on average many errors. -- GreenC 17:45, 16 July 2016 (UTC)

There are 971 cases of example.com in use. Most appear to be non-mainspace. -- GreenC 13:17, 17 July 2016 (UTC)
Yes, it would be nice if the bot excluded example.com links, and/or text inside of <source> that looks like a link (the template that is added doesn't even work there). LiberatorG (talk) 23:51, 17 July 2016 (UTC)
That's going to need some creative thinking on my part. I'll see if I can do something.—cyberpowerChat:Online 07:33, 18 July 2016 (UTC)
I'm working on an implementation now.—cyberpowerChat:Limited Access 18:23, 18 July 2016 (UTC)
 Done IABot is now even more intelligent.—cyberpowerChat:Limited Access 19:08, 18 July 2016 (UTC)

Cyberbot II replacing valid IA archiveurl with one that breaks output of citation template

In this edit on 5 July, Cyberbot II moved an Internet Archive URL from the {{cite web}} "url" parameter to "archiveurl", but in doing so it replaced a valid IA URL already in "archiveurl" with a web.archive.org/web/* URL which causes the citation template to generate an error message and add a hidden category. It also removed a valid "archivedate", which generates another error message.

Original
  • {{cite web |archiveurl=http://web.archive.org/web/20050206040805/http://www.10gea.org/ |archivedate=2005-02-06 |url=http://web.archive.org/web/*/http://www.10gea.org |title=10 Gigabit Ethernet Alliance website}}
  • "10 Gigabit Ethernet Alliance website". Archived from the original on 2005-02-06.
After Cyberbot II edit
  • {{cite web|archiveurl=https://web.archive.org/web/*/http://www.10gea.org |archivedate= |url=http://www.10gea.org |title=10 Gigabit Ethernet Alliance website |deadurl=unfit }}
  • "10 Gigabit Ethernet Alliance website". {{cite web}}: |archive-url= is malformed: timestamp (help); Unknown parameter |deadurl= ignored (|url-status= suggested) (help)
What it should have done
  • {{cite web |archiveurl=https://web.archive.org/web/20050206040805/http://www.10gea.org |archivedate=2005-02-06 |url=http://www.10gea.org |title=10 Gigabit Ethernet Alliance website |deadurl=unfit }}
  • "10 Gigabit Ethernet Alliance website". Archived from the original on 2005-02-06. {{cite web}}: Unknown parameter |deadurl= ignored (|url-status= suggested) (help)

I read through this talk page and its archive back to 5 July, but did not see this particular scenario mentioned. I think it is something that needs to be corrected in the bot, if it has not been fixed already. Thanks. -- Zyxw (talk) 03:07, 20 July 2016 (UTC)

  • Stale
    IABot no longer acknowledges timeless archive URLs as archives.—cyberpowerChat:Online 06:48, 20 July 2016 (UTC)

Around 50 articles got caught by CheckWiki with the bot doing cite templates as *{{|url=http://www.abu2012seoul.com/ |title=Archived copy .... Examples are ABU TV Song Festival 2012, ACES Colombia and Amora London. I don't know how you keep track of all your bots. I have a hard time doing one. My head would explode if I had that many bots. Bgwhite (talk) 06:32, 18 July 2016 (UTC)

Already working on fixing that.—cyberpowerChat:Online 06:59, 18 July 2016 (UTC)
Thanks. I'll cleanup any of the oopses that CheckWiki found. Bgwhite (talk) 07:27, 18 July 2016 (UTC)
You can safely revert them. IABot will come back and do it right when I reactivate it.—cyberpowerChat:Online 07:29, 18 July 2016 (UTC)
This specific bug should be  Fixed now.—cyberpowerChat:Online 07:34, 18 July 2016 (UTC)


Some different cases I've found, Campeonato Brasileiro Série A, East End of London, Gret Palucca, Kushner Companies, Royal Securities Exchange of Bhutan, Utica Psychiatric Center, Uwe Boll, Waltzing Matilda, Wang Shichong, Xen Coffee, Xinjiang conflict, Yossi Vardi and Zardana. I've reverted the bot in all cases. Bgwhite (talk) 19:52, 18 July 2016 (UTC)

 Fixed and tested on all articles.—cyberpowerChat:Online 17:07, 20 July 2016 (UTC)

Can these be added?

I pinged you with my thanks on my talk page but I did just come on this Talk:ABC Daytime/Archives/2016#External links modified. Can these edits that only add a "dead link" tag be adjusted so that they do not show up in the category as well? Thanks for your time. MarnetteD|Talk 17:38, 16 July 2016 (UTC)

 Done—cyberpowerChat:Limited Access 19:43, 18 July 2016 (UTC)
Many thanks. A couple days late :-) MarnetteD|Talk 18:15, 20 July 2016 (UTC)

Thank you

Thank you for creating this extremely usefull both, this was such a wanderfull idea. FkpCascais (talk) 21:29, 20 July 2016 (UTC)

Question: at Talk:Zlatko_Kostić the link from rsssf.com was working fine and it was replaced, why? FkpCascais (talk) 23:24, 20 July 2016 (UTC)
It was flagged as dead in the DB. I've reset the live state. Likely a remnant from the trial.—cyberpowerChat:Limited Access 05:36, 21 July 2016 (UTC)

This edit "fixed" a working link, not the first time that this has happened. Alansohn (talk) 23:38, 20 July 2016 (UTC)

Same her with this link. Alansohn (talk) 00:04, 21 July 2016 (UTC)

I've reset their live states in the DB.—cyberpowerChat:Limited Access 05:39, 21 July 2016 (UTC)

InternetArchiveBot error

The bot appears to have dumped error message page html text into Ziviyeh, Saqqez. --Bamyers99 (talk) 00:03, 21 July 2016 (UTC)

Likewise, the bot did major damage to Craig Stapleton (rugby league). Dl2000 (talk) 00:18, 21 July 2016 (UTC)
It would appear the MW servers are flaking on the bot. If it keeps happening let me know. The seems to be a rare occurrence.—cyberpowerChat:Limited Access 05:41, 21 July 2016 (UTC)

Problem with InternetArchiveBot

Most edits I find are okay, but this one was pretty bad. It seemed to take a URL that was in the "External links" section of the article and apply it to four of the five links it modified. Especially, note the link it modified in the infobox; this is already handled by the infobox code and shouldn't be touched. I hope these problems are fixable. — Gorthian (talk) 00:42, 21 July 2016 (UTC)

That's new.—cyberpowerChat:Limited Access 05:41, 21 July 2016 (UTC)

InternetArchiveBot cosmetic changes

In this edit [4] the only change the bot made was adding spaces between parameters, despite the edit summary of "Rescuing 1 sources and tagging 0 as dead. #IABot (v1.1)" - Evad37 [talk] 03:46, 21 July 2016 (UTC)

Sigh. Not this bug again. :/ — Preceding unsigned comment added by Cyberpower678 (talk • contribs) 05:42, 21 July 2016 (UTC)

date format

Tracked in Phabricator
Task T141210

This bot is currently working its way through the Lexus articles. I like what it is attempting to do. Unfortunately it is inserting 'MMM dd, yyyy' style dates into articles that are already using 'yyyy-mm-dd' style dates in the references. 'yyyy-mm-dd' style dates are a legitimate format in references, therefore this bot is violating MOS:DATERET and MOS:DATEUNIFY. The bot needs to read the existing |date= and |access-date= fields for the reference and use that style. If the reference does not have a date field (sadly, this is still an all too common problem) then the both needs to use the date format from another reference.

Cyberbot II has a similar problem. I spent about 3 months doing almost nothing except cleaning up after it until it had worked its way through all the articles I cared about. That was a very frustrating 3 months that left me rather frazzled.

Once again, I support the task this bot is trying to do and am willing to help out to fix this problem.  Stepho  talk  08:49, 21 July 2016 (UTC)

Sorry, but you're asking sounds unreliable. What if the reference it reads also has a wrong format? This is what the {{usedmy}} and such templates are for. I can make ISO the default however, in the absence of those tags.—cyberpowerChat:Online 09:04, 21 July 2016 (UTC)
The problem is we would need another set of tags because articles may use ISO format for access/archive dates even if they use mdy/dmy for publication dates. nyuszika7h (talk) 09:13, 21 July 2016 (UTC)
And with that we have a level of complexity, that simply can't be reliably programmed in a bot. I'm not going to waste countless developing a way to detect date formats that simply will not work reliably. This isn't me saying I'm not willing to do something about, I'm just being realistic about it.—cyberpowerChat:Online 09:15, 21 July 2016 (UTC)
(edit conflict) Yeah, and it can be easily fixed with MOSNUMscript (or manually) by another editor, it's not like the bot makes that many edits per article, and it already notifies editors on the talk page anyway. There is no reliable way for the bot to detect this currently, so I don't think it's fair to blame the bot. nyuszika7h (talk) 09:17, 21 July 2016 (UTC)
I've re-enabled the bot per this. When reverting the run page edit, for some reason my edit summary disappeared.—cyberpowerChat:Online 10:19, 21 July 2016 (UTC)

|df= is intended to make this sort of problem easier. The bot can write all dates in ymd format and then add an empty |df= parameter. Editors then need only add a keyword to it (|df=dmy or mdy) which will cause Module:Citation/CS1 to render all dates except access and archive dates in the specified format. If {{use dmy dates}} or {{use mdy dates}} is present, the bot can set |df= to the appropriate value. To set all dates including access and archive dates to the same format, use |df=dmy-all (or mdy-all).

—Trappist the monk (talk) 10:40, 21 July 2016 (UTC)

Now that can easily be done. :-)—cyberpowerChat:Online 11:34, 21 July 2016 (UTC)

At Talk:Zendaya § External links modified, the second link is not dead (though it doesn't display the album details and wants to open in iTunes, which may or may not have confused the bot), and the third link is not dead either. (Also, I just noticed it used HTTP instead of HTTPS for the second one.) nyuszika7h (talk) 10:43, 21 July 2016 (UTC)

The iTunes link is indeed coming back as a false positive, I can investigate that, but the second one is also a remnant from the trials. I'm running a purge script right now. Given it's size, I imagine it's going to take at least 24 hours to complete.—cyberpowerChat:Online 12:07, 21 July 2016 (UTC)

Hey, the archived links that InternetArchiveBot are trying to fix to http://factfinder2.census.gov don't work; they produce a System Unavailable error message. Graham87 11:05, 21 July 2016 (UTC)

An example of that is at Oakwood, Georgia. Also, the birdlife.org links, like the one added to Wyndham, Western Australia, don't appear to work either (though that could either be a WebCite or screen reader issue). Graham87 11:13, 21 July 2016 (UTC)
Never mind re Birdlife; it does work, eventually. Graham87 11:18, 21 July 2016 (UTC)
You may need to use {{cbignore}} on the non-working one, or find a better URL that works.—cyberpowerChat:Online 12:10, 21 July 2016 (UTC)

Bot using too old archive

Tracked in Phabricator
Task T141212

At Talk:Dakota Fanning § External links modified 3, the bot used an archived version that is too old and did not yet confirm the claim. It would be probably the best if the bot used the archived copy closest to the access date. nyuszika7h (talk) 10:34, 21 July 2016 (UTC)

It already does that. The newer archives don't work.—cyberpowerChat:Online 10:36, 21 July 2016 (UTC)
I replaced it with with the one directly after the bot's pick, and it does work. nyuszika7h (talk) 10:38, 21 July 2016 (UTC)
I looked to and the one I saw after that was a broken redirect.—cyberpowerChat:Online 11:31, 21 July 2016 (UTC)
Not sure what you're looking at, but this works fine for me. nyuszika7h (talk) 11:35, 21 July 2016 (UTC)
That's not one of the URLs you mentioned, and that archive is years older than the two you mentioned.—cyberpowerChat:Online 12:12, 21 July 2016 (UTC)
I linked to the section dated 2 July 2016, that's what I was talking about. Also, perhaps the bot should consider titling sections "External links modified (2 July 2016)" or similar. It's possible to link to the second, third, etc. section by adding a number like I did, but it would be more clear in the TOC and in edit summaries the auto-generated link will always go to the first section with that title. nyuszika7h (talk) 12:15, 21 July 2016 (UTC)
I really need to add some magic keywords for the confit page.—cyberpowerChat:Online 14:20, 21 July 2016 (UTC)

Unnecessary additions by InternetArchiveBot

Tracked in Phabricator
Task T141213

Hi, just wanted to alert you to the following unnecessary additions by InternetArchiveBot – the archive URLs had already been provided in the citations. — SMUconlaw (talk) 12:49, 21 July 2016 (UTC)

Noted. Will fix in the next release.—cyberpowerChat:Online 14:21, 21 July 2016 (UTC)

This edit is another where a working link is misidentified as dead. Alansohn (talk) 13:14, 21 July 2016 (UTC)

See similar threads above.—cyberpowerChat:Online 14:22, 21 July 2016 (UTC)

In Zlatko Čajkovski this edit is wrong, because http://www.rsssf.com/miscellaneous/double-caps.html works. The same problem occurrred with other URLs from http://www.rsssf.com in Xabi Alonso [5], URL http://www.rsssf.com/miscellaneous/spanpoy.html and Patrice Bernier [6], URL http://www.rsssf.com/miscellaneous/can-recintlp.html --Jaellee (talk) 13:23, 21 July 2016 (UTC)

See similar threads above.—cyberpowerChat:Online 14:26, 21 July 2016 (UTC)

Reports -> Phab

Hi CP678, apologies for another email - I noticed that you're getting a hell of a lot of reports about your bots. I see that a couple of phab tickets were opened on Community Tech regarding it, but was wondering if it'd be worth requesting a phab project be set up for some of your common bots and directing all reports there? People will still leave messages, but I imagine making phab the "official" method will cut down email traffic and allow you to categorise/prioritise? Just my 2c, thanks for all the hard work -- samtar talk or stalk 14:57, 21 July 2016 (UTC)

Yea, I'm thinking that too.—cyberpowerChat:Online 15:14, 21 July 2016 (UTC)

See this edit. There is still nothing wrong with this link. Please stop trying to fix what isn't broken. If the problem will be solved in 24-48 hours, please stop the bot until the problems are resolved. Alansohn (talk) 17:08, 21 July 2016 (UTC)

Reports

It's tagging them as dead.—cyberpowerChat:Limited Access 05:40, 21 July 2016 (UTC)
Are you sure? The dead-url arg is different from {{dead}}. The dead-url arg only tells the template how to display the URL, it's not meant to be a dead link tagger on its own. There's no date mechanism or hidden category or bot ID. That has been my understanding, it's used in conjunction with the archive-url argument for template display purposes. -- GreenC 12:25, 21 July 2016 (UTC)
Whoops. My bad. I'll fix that in the next release.—cyberpowerChat:Online 14:15, 21 July 2016 (UTC)
Would those links get tagged with {{dead}} on re-run post-fix? Leaving the dead-url argument in place shouldn't hurt anything. -- GreenC 14:37, 21 July 2016 (UTC)
It should.—cyberpowerChat:Online 14:52, 21 July 2016 (UTC)

Duplicated entries

Tracked in Phabricator
Task T141214

Can you please stop your bot from making edits like these [7] [8] ones? The only thing it is doing is duplicating the links to Wayback archives, as they are also included in a {{Wayback}} template.--Jetstreamer Talk 17:30, 21 July 2016 (UTC)

Furter to the above posts, I've indeffed the block. Any admin may unblock without further reference to myself. I've noticed a lot of the bot's edits have been reverted by numerous editors. This, plus messages here and at the bot's talk page lead me to conclude that something is not working properly and the best course of action is to temporarily block the bot. Mjroots (talk) 19:26, 21 July 2016 (UTC)
@Mjroots: I fail to see how duplicating links can be called "vandalism", it's undesirable but it's not inherently harmful unlike blanking refs which is something it did earlier. Either way, blocking was not necessary, it can be shut off from its user page and someone else has already done so a few minutes before you blocked it, and it stopped editing. nyuszika7h (talk) 19:34, 21 July 2016 (UTC)
Never mind about the vandalism part, it was another user who called it that on the bot's talk page. nyuszika7h (talk) 19:35, 21 July 2016 (UTC)
@Nyuszika7H: - I made the block in good faith and it is not to be seen as a discredit to the bot owner. This is the first time I've ever had to block a bot, so I appreciate that I may not have got everything right here. I noticed a number of difference editors reverting the bot's edits, plus messages left which led me to conclude that the bot was malfunctioning. The block was made to protect the project as a whole. I'm sure that the problem will be overcome in due course and the block will be lifted. Mjroots (talk) 19:41, 21 July 2016 (UTC)
@Mjroots: Yeah, sorry, as I said the first part doesn't apply as I read it wrong and it was another user who called it vandalism. It was not strictly necessary to block since it has a shutoff page and it was already disabled there a few minutes earlier, but I guess you were not aware of that. nyuszika7h (talk) 19:44, 21 July 2016 (UTC)
A block is not entirely inappropriate. To Cyberpower, I mean this in the nicest way possible, but we should kink out the bugs on a different wiki instead of re-enabling the bot after others have disabled it with good reason — MusikAnimal talk 19:46, 21 July 2016 (UTC)
Agreed.—cyberpowerChat:Limited Access 20:34, 21 July 2016 (UTC)

Tagging 0 as dead

Regarding this edit, your InternetArchiveBot said "... and tagging 0 as dead" although there are many dead URLs on the page (e.g. Haaretz ones').

Just to report so bot might get improved. --Obsuser (talk) 15:58, 21 July 2016 (UTC)

There's more to the engine than you know.—cyberpowerChat:Online 17:49, 21 July 2016 (UTC)
I don't understand what you want to say.
I've just informed you that there are many dead URLs on the page (e.g. Haaretz ones'); maybe 0 are tagged as dead but there are indeed some dead URLs left...--Obsuser (talk) 17:52, 21 July 2016 (UTC)
It takes at least 9 days for the bot to acknowledge them as dead. During that time it runs three checks. If they all fail, they get classified as dead. Otherwise the bot moves on.—cyberpowerChat:Limited Access 20:35, 21 July 2016 (UTC)
Tracked in Phabricator
Task T141217

Sorry to pile on the bad news.

See [Indiana Botanic Gardens, Individualism (look for recollectionbooks), Li Wu and Quanrong. There are about 20 more that I've found and they all follow the same pattern. I will fix them all. Bgwhite (talk) 08:18, 22 July 2016 (UTC)

InternetArchiveBot: Text duplication

Tracked in Phabricator
Task T141219

Yesterday your bot copied 137 kB duplicate text in article Queer as Folk (season 1). Please check your bot before running again! --GünniX (talk) 10:44, 22 July 2016 (UTC)

Hi Cyberpower678, last August Cyberbot II tagged Disumbrationism as containing a blacklisted link (http://reverent.org/bad_painting_contest.html), but I'm unable to find reverent.org on the local or global blacklists. Is it possible that the link was whitelisted since then? I'm not very knowledgeable about how blacklisting and whitelisting work. Thanks, --momoricks 19:36, 20 July 2016 (UTC)

Possibly, yes.—cyberpowerChat:Limited Access 05:33, 21 July 2016 (UTC)
Okay, thanks. I searched for "reverent" and "reverent.org" in the whitelist archives located at MediaWiki talk:Spam-whitelist but came up empty. Can you advise me on where to look for the link's list status? If it is whitelisted, should I contact you to have the tag removed? --momoricks 18:16, 21 July 2016 (UTC)
Aha, I found it here: https://meta.wikimedia.org/wiki/Spam_blacklist. Best, --momoricks 18:05, 22 July 2016 (UTC)

Hello. The bot tagged numerous links to the Inter-Parliamentary Union website as dead today (e.g. this, but they aren't. I guess there may be some kind of error? Cheers, Number 57 20:26, 20 July 2016 (UTC)

It's possibly a remnant from the trial. If you give me a list of links, I can reset their state in the DB.—cyberpowerChat:Limited Access 21:02, 20 July 2016 (UTC)
I'm not sure what you mean by a list of links, but so far every link it's tagged to the http://www.ipu.org/ website (which it's continuing to do) are still live. Cheers, Number 57 11:00, 21 July 2016 (UTC)

I'm notifying you of my comment at Talk:Vera Brosgol#External links modified; InternetArchiveBot (talk · contribs) archived a link properly, but tagged the source as dead when it's not. Just a heads-up. — fourthords | =Λ= | 21:20, 20 July 2016 (UTC)

I reset the link state. It shouldn't happen again.—cyberpowerChat:Limited Access 21:25, 20 July 2016 (UTC)
I just had to revert several bot-edits tagged Rescuing 0 sources and tagging 1 as dead. #IABot (v1.1), concerning a MPC-citation (www.minorplanetcenter.net). The URL is not dead (except for those with a blocked IP due to excessive http-requests). Also, the re-formatting of the citation adding trailering white-spaces is unhelpful, even if the URL were dead. Thx for noticing, Rfassbind – talk 10:48, 21 July 2016 (UTC)
A purge script is currently running to reset the false positives.—cyberpowerChat:Online 14:16, 21 July 2016 (UTC)
The bot is continuing to do the identical same mistake

as I have previously described above. See edits in articles Krisztián Sárneczky, Viktor Knorre and Nobuhiro Kawasato. Thanks for noticing, Rfassbind – talk 19:59, 22 July 2016 (UTC)

@Rfassbind: The bot has been already shut off (and temporarily blocked) since then. nyuszika7h (talk) 20:14, 22 July 2016 (UTC)

InternetArchiveBot marked live source as dead

The RGS site marked dead by this change to the Nikolay Przhevalsky article is actually functioning. I changed the deadurl flag on the ref, but you may want to examine the effectiveness of the bot's algorithm. Rupert Clayton (talk) 03:53, 21 July 2016 (UTC)

The bot's algorithm has been confirmed to have a false positive rate of 0.1%. What you're seeing here are the lingering remnants of the trial from when this algorithm was still being tested, developed, and approved, lurking around in the DB. The algorithm itself confirms them to be alive. It would seem I need to run a purge script to get the lingering remnants cleaned up from the DB.—cyberpowerChat:Limited Access 05:47, 21 July 2016 (UTC)
Thanks for the insight. The bot's edit was pretty recent, but I guess you're saying that the database of URL availability still has some old inaccurate data in it. That purge sounds like a good idea. Rupert Clayton (talk) 23:18, 22 July 2016 (UTC)

Template:Cratstats

Hi Cyberpower678, would you check Cyberbot_I's config for updating Template:Cratstats - I think it has an off-by-one bug in the BRFA section. Thank you, — xaosflux Talk 04:18, 23 July 2016 (UTC)

X! Tools not working

Could you fix that please? Softlavender (talk) 03:36, 23 July 2016 (UTC)

Seem to be fixed now. Softlavender (talk) 04:51, 23 July 2016 (UTC)

Problem with reference

Hi, there appears to be some problem with this change made by BOT, could be something to do with embedded ref in {{Official website}} template. Keith D (talk) 20:46, 22 July 2016 (UTC)

Another example. Keith D (talk) 21:15, 22 July 2016 (UTC)
(edit conflict) @Keith D: Well, {{Official website}} is not supposed to be used in references, so the articles should be fixed either way, and then the bot won't break on it. nyuszika7h (talk) 21:22, 22 July 2016 (UTC)
I'll note that in those cases, the link was not actually dead, but I assume that will be fixed by the database purge. nyuszika7h (talk) 21:25, 22 July 2016 (UTC)
I have just reverted them for now. Wonder how many more there are? Keith D (talk) 21:30, 22 July 2016 (UTC)
@Keith D: I have fixed the references on those two articles. Not sure how many but this particular issue shouldn't be that common (they were likely added by the same user), but anyway the bot has been temporarily shut off (and blocked) for now until some issues are fixed. nyuszika7h (talk) 21:52, 22 July 2016 (UTC)
Sorry, but there's nothing I can do here.—cyberpowerChat:Limited Access 18:58, 24 July 2016 (UTC)

alternate |dead-url= keyword

Continues from these discussions: |dead-url=unfit and |dead-url=unfit maintenance category.

Unless the bot knows without a doubt that a url is unfit, it should not make that assertion. I have added an alternate, bot-specific keyword to Module:Citation/CS1/sandbox that, to the casual reader and editor, acts the same as |dead-url=unfit. It differs in how the citation is categorized. The new keyword will become part of the live module suite over the weekend of 30–31 July.

The bot-specific keyword is bot: unknown. Please use that instead of unfit.

—Trappist the monk (talk) 11:00, 23 July 2016 (UTC)

Will be implemented in the next release. Thank you. :-)—cyberpowerChat:Online 17:47, 23 July 2016 (UTC)
 Done—cyberpowerChat:Online 10:19, 25 July 2016 (UTC)

WebCite URL problem

Here, the bot incorrectly added an archive URL as "https://web.http://www.webcitation.org/5hRn3DyEf". – nyuszika7h (talk) 16:22, 25 July 2016 (UTC)

Also a lingering remnant of when the bot only supported archive.org archives. Another more heavy purge script may be needed.—cyberpowerChat:Online 10:08, 26 July 2016 (UTC)

Just some more data

While you are working on getting the InternetArchiveBot spruced up, I wanted to give you some more input on its behavior that I don't think I've seen addressed here (though I may have missed it). Here are some specific problems pulled from my watchlist over the past few weeks:

  1. Global warming, July 2 (yes, it's old, but it could still be good data...): In this edit, the edit summary said "rescuing 7 sources and tagging 0 as dead", but the actual edit modified only two citations and marked them both as dead. On the talk page, however, all seven links are listed, including one that isn't even in the article. (The edit also removed spaces before parameters but added them after.)
    This isn't really a bug, sometimes, if it can't correctly fix the source, it won't but it will still mention it on the talk page. Though I can investigate why they aren't getting fixed.—cyberpowerChat:Online 09:19, 25 July 2016 (UTC)
    I would guess that's because the citations in that article are all over the place, combining {{cite}}, {{citation}}, {{harvnb}}, {{DOI}}, bare links, the kitchen sink, and tacked-on text with links. All the links (except one) listed on the talk page, however, are all working. Not sure why it would choose just those.— Gorthian (talk) 18:00, 25 July 2016 (UTC)
    What I mean is, sometimes it can't fix the wikimarkup, so it only gets mentioned on the talk page.—cyberpowerChat:Online 10:02, 26 July 2016 (UTC)
  2. Yosemite National Park, July 16: This edit mangled a perfectly functioning {{Wayback}} template in "External links" and added the parameter "deadurl=unfit".
    This was fixed in the v1.1 release.—cyberpowerChat:Online 09:19, 25 July 2016 (UTC)
  3. Yosemite National Park, July 20: I detailed the problems with this edit here, and you remarked that it was a new one.
    I forgot to create a ticket for that one. Too many bug reports. Filed under T141248.—cyberpowerChat:Online 09:19, 25 July 2016 (UTC)
  4. Geology (edit) and Quick clay (here), July 21: The bot inserted the {{Wayback}} template into references. What happened to having it use a cite template (for bare URLs without text) and leaving it alone if there is text? It used to be able to do that.
    There's nothing wrong with those edits. Because the links have text it adds a wayback tag to it, and leaves the link alone.—cyberpowerChat:Online 09:19, 25 July 2016 (UTC)
Putting the Wayback template into a citation makes a mess of the citation and makes it a real chore to clean up. What GreenC said in the last discussion about this was "there are other tools to semi-automatically convert bare links to cite web + archive URL. However if the Wayback template is inserted by Cyberbot, those tools won't work and it must be done 100% manual. So it's actually creating more work for someone to later clean up." Maybe using the Wayback template isn't the best solution in citations. There must be another way to insert an archive link into a bare-link reference. — Gorthian (talk) 18:00, 25 July 2016 (UTC)
Three types of refs a bot could ideally convert to a cite template:
  1. http://site.com
  2. [http://site.com]
  3. [http://site.com title]
If the ref contains other text fallback to wayback template (see no other option). I believe CB is doing 1 & 2? -- GreenC 18:45, 25 July 2016 (UTC)
It's the ones with extra text I'm talking about. Wouldn't it be possible to insert a second "bare URL" immediately after the original that would contain the Wayback URL and date? I'm going to have to go dig up an example... — Gorthian (talk) 19:00, 25 July 2016 (UTC)
Okay, here's the original reference.[1] And here's the archived reference.[2] Wouldn't this be do-able? — Gorthian (talk) 19:14, 25 July 2016 (UTC)
If that was done it should be templated to keep maintaince of links cleaner for bots and people. In which case why not keep using {{wayback}}? If it's a display concern, which is understandable, we could patch the wayback template to have an "inline" switch to change how its displayed to be more suitable for inline citations (the template was originally designed for external link sections in mind not for use in citations). -- GreenC 02:20, 26 July 2016 (UTC)
Yes, display is a problem (I especially dislike the "Wayback Machine" link intruded into the citation), but the way the date parameter is formatted (e.g., |date=20050107123045) is tedious to unravel manually, and other tools don't know what to do with it. Plus, the Wayback Machine isn't the only archive possible. Maybe a new template is in order? — Gorthian (talk) 16:02, 27 July 2016 (UTC) (apologies for the delay in replying)
A new template may be a solution. Using the wayback template inside citations was probably never a good idea but evolved organically before Cyberbot to address this problem. If the new template was in Lua it could automatically detect the date from the URL and, the archive name, so would only require the URL which would be easy shortcut. If the URL was short-form like archive.is or webcite it could have optional arguments for the date in mdy or dmy or 14-digit. It would also have a bot arg to keep track of who added it. Ping @Trappist the monk: any other ideas or concerns. -- GreenC 17:06, 27 July 2016 (UTC)
I don't think that I have never used or edited {{wayback}}. Why am I being pinged?
—Trappist the monk (talk) 21:30, 27 July 2016 (UTC)
The conversation deals with displaying citations. I thought you might have some other insight I was missing but no trouble if not. -- GreenC 22:44, 27 July 2016 (UTC)

References

  1. ^ "Family dead in basement after sinkhole ate home". CNN, May 12, 2010.
  2. ^ "Family dead in basement after sinkhole ate home"(archived copy from May 15, 2010). CNN, May 12, 2010.

Hope the revamp is chugging along. Thanks. — Gorthian (talk) 05:43, 25 July 2016 (UTC)

I haven't had much time the last few days, but I'm going to start hammering away these bugs.—cyberpowerChat:Online 09:19, 25 July 2016 (UTC)

Note about weird (but insignificant) behavior

Hello. Just thought you might like to know that your bot created an empty page, trying to remove pp-pc1 from a page that was deleted 6 minutes earlier: Abel Tesfaye (the Weeknd). I've speedied the page and just thought I'd let you know. Cheers ~ | twsx | talkcont | ~ 19:40, 27 July 2016 (UTC)

It was probably running through its list during that time.—cyberpowerChat:Online 07:14, 29 July 2016 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=User_talk:Cyberpower678/Archive_36&oldid=1084106141"