User talk:cyberpower678

This user is offline, or has forgotten to update this message since starting a wikisession.

_{(If there have been multiple edits from this user in the last 60 minutes and the most recent one wasn't to activate this template, it is safe to assume that this user forgot.)}

Senior Editor II

—CYBERPOWER (Around)
Find out why my signature changes color.

Hello!! I am Cyberpower678. I am an administrator on Wikipedia. Despite that, I'm still your run of the mill user here on Wikipedia.
I specialize in bot work and tools, but I lurk around RfPP, AfD, AIV, and AN/I, as well as RfA. If you have any questions in those areas, please feel free to ask. :-)
For InternetArchiveBot matters specifically, please see meta:InternetArchiveBot/Problem for common issues or meta:User talk:InternetArchiveBot to leave a message
I also serve as a mailing list moderator and account creator over at the Account Creation Center. If you have any questions regarding an account I created for you, or the process itself, feel free to email the WP:ACC team or me personally.
At current I have helped to create accounts for 2512 different users and renamed 793 other users.
Disputes or discussions that appear to have ended or is disputed will be archived.

All the best.—cyberpower

View my talk page Archives.

Requests for adminship and bureaucratship update
RfA candidate	S	O	N	S %	Status	Ending (UTC)	Time left	Dups?	Report
Worm That Turned	261	4	7	98	Open	09:47, 18 November 2024	1 day, 18 hours	no	report

IABot deactivated on nlwiki

Hello, the nlwiki community reported that IABot was malfunctioning. Way too many live links were marked as dead; the decision was taken to deactivate the bot on nlwiki. I did some investigating and discovered a possible cause: Spamhaus' CBL has blacklisted (what I presume to be) the IP address used by the bot, see here. Is this plausible? --bdijkstra (talk) 17:53, 15 June 2020 (UTC)[reply]

Bdijkstra, that is the IP of the bot, but I wonder why it got blacklisted. I mean yea, it's a bot, but it's not malicious. I have removed the listing. Maybe that may help resolve things. —CYBERPOWER (Chat) 18:14, 15 June 2020 (UTC)[reply]

Looking at the description it is most likely not the bot software that is the problem, but some botnet program running on the same machine or on a different machine behind the same NAT router. --bdijkstra (talk) 18:22, 15 June 2020 (UTC)[reply]

Bdijkstra, Nope. This IP belongs to IABot only. —CYBERPOWER (Chat) 18:23, 15 June 2020 (UTC)[reply]

I hope you're right. Is there a way to tell IABot to only check the live status of an URL? I'd like to check a few links before enabling the bot again. --bdijkstra (talk) 08:29, 16 June 2020 (UTC)[reply]

I guess you missed my question? I didn't see it in the FAQ. --bdijkstra (talk) 08:19, 19 June 2020 (UTC)[reply]

Bdijkstra, I am however exploring methods of setting certain IABot jobs to appear as different IPs based on whatever region Wikipedia it's working on. This is in hopes of getting around some geo-restrictions. So on nlwiki, IABot will appear to third party sites as making requests from the Netherlands. But a setup like this is insanely difficult to setup.—CYBERPOWER (Chat) 18:51, 15 June 2020 (UTC)[reply]

Shouldn't it be based on the geo-location where the site of the reference (URL) resides? References to Dutch sites, not only exist on nlwiki. You don't bypass geo-restrictions on cnwiki by making it appear as a request from China. -- Sb008 (talk) 22:55, 15 June 2020 (UTC)[reply]

Sb008, You're overthinking it. A certain language Wikipedia is not likely going to include a link that is geo-blocked for them. Enwiki is least likely to include a Japanese geo-blocked site, and vice versa. —CYBERPOWER (Chat) 23:02, 15 June 2020 (UTC)[reply]

I would not have thought that IP geolocation is an important reason to block IABot. Anyway, perhaps there are VPN providers that allow dynamic programmatic country selection? --bdijkstra (talk) 08:29, 16 June 2020 (UTC)[reply]

Bdijkstra, I have a VPN provider that lets me move my digital presence at will. I'm going to see if I can integrate that. —CYBERPOWER (Around) 12:24, 16 June 2020 (UTC)[reply]

Admin stats

The bot seems to have stopped updating AdminStats before it got to mine. That issue aside, I'm not sure the bot controls this, but I've noticed over the last several days that it wasn't updating exactly accurately. Don't know if you or your bot is what updates the xtools.wmflabs.org, but that's been off for a few days. Specifically, what I have deleted at at CSD has not been even closely accurate. I think this is a recent thing. If I delete 30 pages, it might log that I deleted 5 or 10. It's been under-counting those stats for several days. Haven't compared it with any of the other stats. — Maile (talk) 18:42, 30 July 2020 (UTC)[reply]

Well, maybe this is the cause: Toolserver replication lag for enwiki — Maile (talk) 23:48, 30 July 2020 (UTC)[reply]

Admin stats are back up and running. — Maile (talk) 12:33, 2 August 2020 (UTC)[reply]

Changing Username

Current Username - Prashantballal Change to - Prashant Ballal

I just joined Wikipedia and madhe a mistake creating username. So I request you to make changes as mentioned above. Prashantballal (talk) 09:36, 31 July 2020 (UTC)[reply]

@Prashantballal: Why are you asking here? Follow the instructions at WP:CHU. (talk page stalker) Jackmcbarn (talk) 20:59, 31 July 2020 (UTC)[reply]

Probably, because Prashant Ballal didn't knew about that page already but felt reasonably encouraged to ask an admin personally by this comment on Cyberpower's user page:

"At current I have helped to create accounts for 2512 different users and renamed 793 other users."

Just guessing...

--Matthiaspaul (talk) 10:38, 2 August 2020 (UTC)[reply]

Administrators' newsletter – August 2020

News and updates for administrators from the past month (July 2020).

Administrator changes

Red Phoenix

Euryalus • SQL

Jujutacular • Monty845 • Rettetast • Madchester

Oversight changes

GB fan

Keegan • Opabinia regalis • Premeditated Chaos

Guideline and policy news

There is an open request for comment to decide whether to increase the minimum duration a sanction discussion has to remain open (currently 24 hours).
Speedy deletion criterion T2 (template that misrepresents established policy) has been repealed following a request for comment.
Speedy deletion criterion X2 (pages created by the content translation tool) has been repealed following a discussion.
There is a proposal to restrict proposed deletion to confirmed users.

Sent by MediaWiki message delivery (talk) 08:20, 1 August 2020 (UTC)[reply]

IA Bot on SqQuote

Hello, Cyber!

Can the IA Bot work on other wiki projects except Wikipedias? I'd like to set it up for work at the Albanian Wikiquote. Do you need anything provided by me? I believe the configurations and translations would need to be like they are for SqWiki (we may have to create 1-2 templates but we'll just import them from SqWiki). - Klein Muçi (talk) 10:31, 1 August 2020 (UTC)[reply]

Many many false positives

Hi! On the Hungarian wiki, InternetArchiveBot is archiving way too many live links. For each dead link, it archives about 10 live links. For instance here or here. Yes I know I could report the links one by one, but even if I report 100 links, the next day 100 others will be wrongly archived. This has been going on for years. Can you please throw an eye on it? Thanks and kind regards, Whitepixels (talk) 07:42, 28 July 2020 (UTC)[reply]

Whitepixels, I've disabled dead link checking on your Wikipedia for now, until this can be more closely investigated. It will only handle tagged dead links. —CYBERPOWER (Around) 12:24, 28 July 2020 (UTC)[reply]

Cyberpower678: I expect this is why IABot is now writing edit summaries on enwiki in Hungarian (based on the above sqwiki thread)? stwalkerster (talk) 13:48, 28 July 2020 (UTC)[reply]

Stwalkerster, yep. :-( I shut down the web interface to prevent further use until this is fixed. —CYBERPOWER (Chat) 14:50, 28 July 2020 (UTC)[reply]

Hi and many thanks for looking into it. Some more examples (if you need any), where (nearly) all archived links were/are live: 1, 2, 3 (here 2 links are dead, but 7 were archived), 4 etc. Kind regards, Whitepixels (talk) 14:52, 28 July 2020 (UTC)[reply]

As it is not clear (to me) from the thread if this is already known, but the English Wikipedia is affected as well. IABot is changing many url-status' to dead which are perfectly fine and live. Just some examples I ran into: [1] [2][3]. Unfortunately, I cannot log in to your tool, therefore, if these URL need to be individually reported, please do it for me. However, as this appears to be a more general problem, reporting individual URLs is probably (hopefully) not necessary, or is it?

BTW. As can be seen in the examples, IABot still messes up the date format. Please adhere to the |cs1-dates=y parameter of the {{Use dmy}}/{{Use mdy}} etc. templates. It is quite annoying when all references were carefully set up to use the ymd format, and your bot comes around and messes them up... (Why does the bot edit the date anyway when the correct date was already present?)

Thanks. --Matthiaspaul (talk) 22:32, 28 July 2020 (UTC)[reply]

Matthiaspaul, IABot follows template formatting directives on articles. When a Use (whatever) dates template exists it overrides IABot's judgement about which date format to use. —CYBERPOWER (Chat) 22:38, 28 July 2020 (UTC)[reply]

(edit-conflict) Thanks for the answer, but IABot does (unfortunately) not adhere to the directive. If f.e. an article contains {{Use dmy dates|date=July 2020|cs1-dates=y}} this means that the dmy format is to be used in the article text and the ymd format is to be used in the references (which is perfectly fine per MOS and many article use this format). Since IABot only deals with references, it is actually the argument given to the |cs1-dates= parameter that is important to the bot (y here, but other values are possible as well, see help), not if the template is named {{Use dmy dates}} or {{Use mdy dates}}. Only, when the |cs1-dates= parameter is not present, IABot should fall back to dmy or mdy depending on the name of the "Use xyz" template ({{Use dmy dates}} or {{Use mdy dates}}). Exception: Individual |df= parameters in specific templates override the format for that specific template. IABot, however, still appears to completely ignore the |cs1-dates= parameter which was introduced about 1.5 years ago. --Matthiaspaul (talk) 23:00, 28 July 2020 (UTC)[reply]

No one informed me of such a change, to the template. But regardless, it seems like a simple config change to the bot should resolve this. But I need to know everything first. What other directives are there?—CYBERPOWER (Around) 23:04, 28 July 2020 (UTC)[reply]

Yeah, I think it should be quite easy to read that parameter. So far, there is only |cs1-dates= which can take nine possible arguments (see Template:Use dmy dates#Auto-formatting_citation_template_dates - also supported by {{Use mdy dates}} and all aliases). --Matthiaspaul (talk) 23:21, 28 July 2020 (UTC)[reply]

If IABot does not support the short date formats already, which are selectable by this parameter, I think it would be reasonably safe to substitute them by their long equivalents (at least much better than not adhering to the parameter at all).

Regarding the |df= parameter of the CS1/CS2 citation templates, this is much less important than the |cs1-dates= parameter of the Use xyz templates (because it only applies to individual citations), but if you'd want to implement support for this as well, I would be willing to research the valid arguments tomorrow (strangely enough, I couldn't find a help page for this right now, only old threads while it was still in development in 2015 and there still were several changes). --Matthiaspaul (talk) 00:11, 29 July 2020 (UTC)[reply]

As IABot continues to incorrectly change the |url-status= of links in citations from "live" to "dead" even for links which are accessible without any problems (f.e. here) after several users have reported the issue, it is high time to either fix the problem or stop the bot also in the English Wikipedia.

Since threads get archived way too fast on this page for a meaningful discussion (less than 3 days), I have restored a related thread and set it to "don't archive" until fixed. Please consider to change the archiving default to something more reasonable like at least two weeks.

Do you need any further input from me in regard to the date format issue with the |cs1-dates= and/or |df= parameters?

--Matthiaspaul (talk) 18:51, 1 August 2020 (UTC)[reply]

Unfortunately, IABot continues to create damage to the project ([4]), and as the problem is not local to a specific url or article, there is no point to block it locally. Therefore: Please stop the bot on the English Wikipedia until the problem is fixed.

Thanks.

--Matthiaspaul (talk) 09:54, 2 August 2020 (UTC)[reply]

As IABot continues to create damage to the project, request to stop IABot until the problem is fixed

Hi Cyberpower, since IABot continues to incorrectly change |url-status= from "live" to "dead" in thousands of citations after several users have reported the problem to you (see above threads, some of them already archived), and as I cannot login to your tool to stop the bot myself, I am now asking you to stop the bot on the English Wikipedia until the problem is fixed. As the problem seems to be present also in other WPs it might be wise to stop it globally. This request is with some regret because IABot is one of the few bots which usually does good work, but the damage it is creating to the project right now is no longer tolerable. As its maintainer you are responsible for the bot's edits. Thanks. --Matthiaspaul (talk) 09:57, 2 August 2020 (UTC)[reply]

and as I cannot login to your tool to stop the bot myself speaking of, shouldn't the kill switch be onwiki? ProcrastinatingReader (talk) 10:08, 2 August 2020 (UTC)[reply]

I agree. So far, it was my impression that a bot's shutoff button must be on-wiki. When I tried to stop the bot myself an hour ago I was perplexed to find out that it is located elsewhere (where I don't have access). All Wikipedians should have access to the shutoff button, otherwise it doesn't make much sense... (To a lesser extent this also holds true for the bug reporting tool, but at least it is possible to discuss problems on Cyberpower's talk page - if the threads would not be archived away after less than 3 days (unless users add {{subst:DNAU}} to keep Sigmabot from archiving)...) --Matthiaspaul (talk) 10:20, 2 August 2020 (UTC)[reply]

I've disabled the bot using the UI. The edits you linked above seem possibly problematic. There's also [5][6] at a glance of recent contribs of the bot. The first link on the first diff isn't dead, and I'm also not sure why the bot wrapped it in a CS1 template rather than using {{Webarchive}}? The second also isn't dead. In either case, I think caution is better than letting the bot run amok, if there are indeed issues. And yes, I do think the shut-off switches should be onwiki - an editor shouldn't have to OAuth into a foreign website (which itself could be down / have issues) to disable the bot. ProcrastinatingReader (talk) 10:42, 2 August 2020 (UTC)[reply]

ProcrastinatingReader, bots are not even required to have run pages. IABot having one is a courtesy and keeping it on the UI provides better oversight of what’s running an not. Every user with an account has access to the UI. —CYBERPOWER (Message) 11:24, 2 August 2020 (UTC)[reply]

I know, I just feel like it'd be a good idea for a high profile and complex bot like this one. Bugs can creep into any software, sometimes for the silliest and smallest reasons, and I just feel making the shut-off switch accessible is a good idea for the odd mishaps that do happen. Just a request, not a demand :) ProcrastinatingReader (talk) 11:30, 2 August 2020 (UTC)[reply]

Matthiaspaul, please explain why you do not have access to the bot UI. If you have an account here, you have access there. —CYBERPOWER (Message) 11:10, 2 August 2020 (UTC)[reply]

Apparently not. I don't have an explanation, but I can't log in there. I already mentioned further above, I also can't log into the bug reporting tool nor into Phabricator (at least the latter requires a separate account AFAIK not covered by the global account already). I also can no longer log into Wikidata, Commons, and a number of other sister projects (at least this once worked years ago), although other language entities still work fine. (Strangely enough, when renaming pages in the English Wikipedia, this can still be seen under my account in Wikidata, whereas manual login doesn't work.) Perhaps this has to do with my (rather strong) security settings, but then I can't change them easily (and also can't risk to fiddle with them at present). At least it is good to know that I should be able to log in to the shutdown button (so it makes sense to investigate this further when circumstances allow for it). Either case, if I am facing this, there are probably other users seeing this as well - so, it might be worth thinking about how to more tightly integrate the shut-off button into the English Wikipedia (similar to how its done for f.e. Sigmabot), so that in a case of emergency it really can be accessed by anyone under any circumstances.

--Matthiaspaul (talk) 11:58, 2 August 2020 (UTC)[reply]

Matthiaspaul, actually you literally are the only one reporting this. But this does make me curious. When you click the login button on the UI, what do you see? Screenshots are preferred, but not required. It sounds like you are suffering from bad session data. Did you try clearing your cookies and logging in again. Usually fixes a lot of problems. —CYBERPOWER (Around) 12:06, 2 August 2020 (UTC)[reply]

Matthiaspaul, Also the run page is not designed to be access by ANYONE. It's designed to be accessed by trusted users. Since putting the run page onto my JS is not a good option, I could technically EC protect it. But that leaves the other Wikipedias. IABot's run pages have seen significant IP abuse in the past. Keeping it on the UI, eliminates this problem entirely. —CYBERPOWER (Around) 12:08, 2 August 2020 (UTC)[reply]

Matthiaspaul, yes, it appears this has gotten too far out of hand, so I will be disabling link checking, globabally when I get to my computer. To add insult to injury, it is not a problem I can reproduce for debugging, so I have been having a really hard time figuring out what’s going on. —CYBERPOWER (Message) 11:22, 2 August 2020 (UTC)[reply]

Thank you. I guess, it already does this, but in order to make the bot immune against temporary glitches with a site or the net, does IAbot probe a link several times over a long enough period of perhaps half a year before changing its status to "dead"? --Matthiaspaul (talk) 11:58, 2 August 2020 (UTC)[reply]

Matthiaspaul, yes. IABot has 7 day waiting period for live links before checking. If a link is suspected to be dead, it must fail 3 consecutive checks, with each check having at least a 3 day waiting period. —CYBERPOWER (Around) 12:03, 2 August 2020 (UTC)[reply]

I can think of many reasons why a site might go off for some while without being dead forever (in these strange times even more than before). Perhaps the waiting period should be extended to 3 or even 6 months to be on the safer side. After all, the bot won't switch a link back to "live" if it would go online again months later, or will it?

However, this is unlikely the problem we are seeing right now, because then it would affect (a few) individual sites, not the many sites of diverse orgins as reported by various users. This appears to be more general in nature, that is, rooted in the bot's implementation somehow rather than the web sites.

As someone reported toolserver time lack issues above, and timing issues are often difficult to track down, another very wild guess (being totally ignorant on the actual implementation): If there are several modules/processes involved, does the system (perhaps only under some further trigger conditions to be found) accidently default to assume "dead" (rather than "live") when the inter-process communication breaks down/times out for some reason thereby creating incorrect "dead" probe events?

Good luck.

--Matthiaspaul (talk) 12:44, 2 August 2020 (UTC)[reply]

Matthiaspaul, If a site goes down for more than 9 days, it's generally got a reliability problem to begin with and an archive should be attached to the reference. So extending the waiting period is not going to solve anything. The bot generally doesn't check links it thinks are dead because, why? But given the issues it's been having, it might be a necessity at this point. I've been wanting to deamonize the bot, so having a dedicated task to check links, especially dead might be a good approach to mitigating this issue.

IABot generally keeps inter-processes to a minimum. You can see the algorithm being used to test dead links at deadlinkChecker on GitHub —CYBERPOWER (Around) 14:44, 2 August 2020 (UTC)[reply]

The bot design was working well enough for years it's probably not the core design idea. One solution is as much logging as possible - log the entire header for every network connection. Then for the false positives, review the headers. So you can isolate if it's external to the bot or internal. -- GreenC 14:58, 2 August 2020 (UTC)[reply]

Another idea for testing purposes is different programs like wget and w3m, in my experience they sometime work on sites that don't work with curl (for no explainable reason). Some header commands:

wget -SO- -q --retry-connrefused --waitretry=5 --read-timeout=2 --timeout=5 --tries=2 --no-dns-cache --no-check-certificate --user-agent="Your Agent String" "https://example.com" 2>&1 >/dev/null

w3m -header_dump "https://example.com"

-- GreenC 15:08, 2 August 2020 (UTC)[reply]

Source code for Cyberbot 1

I would like to know whether you can share with me, the source code of this page, so that I can implement the same in Malayalam Wikipedia. Adithyak1997 (talk) 03:02, 4 August 2020 (UTC)[reply]