User talk:TheSandDoctor/Archives/2018/January

References for articles on US Senate elections, 2020

The original United States Senate election in Oklahoma, 2020 article I created in 2016 (likewise for New Hampshire and South Carolina) were redirects to the general article on United States Senate elections in 2020 (as were most of the articles by state for 2020). Someone has replaced Ithem with separate articles for Oklahoma etc, and they should provide the references! Hugo999 (talk) 09:56, 5 January 2018 (UTC)

Hi there Hugo999, I did not mean to imply that you should be the only one to add the references. You are correct that the burden is that of the other user, Westroopnerd (and other editors willing to pitch in, should they choose). I wish that the page curation tool had a feature to leave the messages to another user, not just the original creator. That said, my apologies for leaving them in the incorrect place, but the rest still stands (thank you for creating the redirect in the first place ). At the time, I also added the {{unsourced}} template to the articles in question. Sorry for the confusion and happy editing! --TheSandDoctor (talk) 16:09, 5 January 2018 (UTC)

Westroopnerd here. I'm mostly just working on having a working "template" page in place for these upcoming elections. The problem is that these templates are just that, templates. Since these elections aren't taking place until the next cycle, there are going to be very few references floating about to use for these pages, and not really a lot of information that can be necessarily cited, period. I'll keep an eye out and add any references that I find, but I wouldn't want to add unrelated or barely relevant references just for the sake of having them. Apologies to Hugo999 for causing headaches as well. Thank you for the barnstar, by the way! Westroopnerd (talk) 10:37, 6 January 2018 (UTC)

Closure of Wikipedia:Articles for deletion/Arab Media Watch

Much as I agree with keeping this, I don't think that it was a good idea to perform a non-admin closure here. It is pretty well inevitable that it will be pointed out that this was a controversial closure, so should only be done by an admin, and that will inevitably lead to much discussion that is irrelevant to the notability of the subject, probably including accusations against both you and me that we are not acting neutrally. Maybe I am wrong and this will all die down quietly, in which case there is no need to undo your close, but I somehow doubt it. 86.17.222.157 (talk) 19:15, 9 January 2018 (UTC)

@86.17.222.157: While I do not believe that it was very controversial and believe that I was acting neutrally as I had never participated with the subject matter, discussion etc., I have reverted the close in favour of further discussion. Further discussion is always an option I am willing to consider. I will watch how this discussion plays out closely (watch listed now). --All the best, TheSandDoctor (talk) 19:24, 9 January 2018 (UTC)

@86.17.222.157: Spartaz closed the discussion as keep. --TheSandDoctor (talk) 07:15, 10 January 2018 (UTC)

RfD

Hi and thank you for helping out at RfD. This is appreciated and I hope you continue to do so. I would just like to note that it will normally be a good idea to only close discussions that have been open for the full period of seven days. The expectation is that a discussion shouldn't be closed early unless a lot of people have already commented expressing the same opinion (WP:SNOW), or the redirect has been legitimately speedy deleted in the mean time. Of course, your explicit willingness to reopen the discussion if asked to will probably mitigate most possible perceptions of unfairness or lack of due process, but this still leaves at least one reason to keep it open: even in discussions with obvious outcomes it sometimes happens that towards the end someone will come up with a new finding that turns the whole thing around. Allowing discussions to run their course leading to the best decision is preferable: there's nothing urgent at RfD. Well, that's just my opinion, I hope I'm not being too fussy? I wonder what other people think: pinging Tavix for a second opinion.

Again, this is only about the timing, I don't have any issues with the closes themselves (I've looked at the ones from 6 Jan and they're good). Even if you agree with me about the timing, there's no need to revert anything: this was just about doing things going forward from here. Your help at RfD is appreciated! – Uanfala (talk) 14:08, 11 January 2018 (UTC)

Thanks for the ping, Uanfala, and I agree with your sentiments completely. I have seen plenty of discussions that look headed for an "obvious" result, only to flip once more information has come available. Wikipedia:Redirects for discussion/Log/2018 January 10#Alt-right haircut is a good example of that. If it was the norm to close discussions early that are headed for a certain outcome, those redirects would have been deleted and possibly no one would even care enough to approach the closer to re-open it. As a matter of personal preference, I usually don't even look at discussions to close them until they hit the log from eight days ago, just so I can be absolutely sure they have gone the full week. To use Alt-right haircut again, I was thinking about opining in that discussion. If I had any idea it would have been closed so soon after the relist, I probably would have left one a comment already (I hadn't because I wanted to have some time to analyse sources). -- Tavix ^(talk) 15:02, 11 January 2018 (UTC)

Sorry for the delay in my response. Uanfala (cc@Tavix:) I agree with you entirely, I just believe there to potentially be possible exceptions to that (for instance, the Ryzen redirect). With that said, I will cease going through the ones that have not elapsed. I was just attempting to help the project and reduce the clutter/backlog and save administrators and others time. Tavix do you want the Alt-right haircut one reversed? I will happily reopen it (or any others) if you wish. Also, what do you mean by "left one already"? Lastly, I shall take your advice about waiting 7-8 days before going through (I assume my jan 3rd closures were okay?). --TheSandDoctor (talk) 21:01, 11 January 2018 (UTC)

I think it'd be best to have the alt-right haircut one reopened, along with any others that haven't reached the full seven days and the result isn't obvious (yes, January 3rd is fine). I'll leave it to you to figure out which one(s) that would be. It's a noble task to help keep the backlog at bay and I appreciate your efforts, but it shouldn't happen at the expense of stifling ongoing discussion. I've clarified my comment above, my apologies for the vagueness. -- Tavix ^(talk) 21:44, 11 January 2018 (UTC)

@Tavix: I did not intended to stifle discussions. I have gone ahead and reverted most of the closes. I will re-evaluate the other closes when I have the time, but unfortunately have to run at the moment. I will keep an eye on all reopened ones to see how they close for learning purposes. Would it be appropriate to re-close them once their (respective) 7 days are up (ie day 8 or 9) if they have not already been closed and the consensus remains as it currently is/within my technical ability, or would that best be left to someone else? --TheSandDoctor (talk) 22:26, 11 January 2018 (UTC)

Uanfala and myself only objected to the timing of the closes, not the actual closes themselves, so I don't think anyone would object if you reclose at the appropriate time (keeping WP:BADNAC in mind of course). -- Tavix ^(talk) 23:24, 11 January 2018 (UTC)

Mangled RfD closure

Something went wrong with this edit of yours. I think I've fixed it, but just a heads-up in case there's a wider problem. PamD 09:06, 18 January 2018 (UTC)

Thanks for bringing that up PamD, that is strange. I do understand that I am responsible for edits that the script makes on my behalf, so I do apologize for that. With that said, I am tagging Evad37 (script creator) so that they are aware of its screw up. This is weird. I will keep a closer eye on my closures & how the script acts. --TheSandDoctor (talk) 23:45, 18 January 2018 (UTC)

Bot issues

TweetCiteBot

Hi, there is a problem with TweetCiteBot edits that is adding the edits to Category:CS1 errors: dates. It appears to be loosing the full |accessdate= specified and just outputting the year as per this and probably the other 35 edits that caused a similar problem. Keith D (talk) 11:38, 27 January 2018 (UTC)

Double dead

It processed the article twice leaving two {{dead link}}. [1][2] It can check for a pre-existing maybe. -- GreenC 04:03, 30 January 2018 (UTC)

Hi there Keith D & GreenC, sorry for the delay in my response(s). I apologize that it has added to that category, the issue causing this is one to do with its REGEX, which I am currently looking into resolving but I am also looking to dissect Cyberpower678's parser to see if that would be a better solution. In the mean time, the bot won't run on here unless I feel I have a fix. As for what GreenC mentioned, that issue shouldn't have happened, but it did and again, I am redesigning the bot. Sorry for the disappointment. --TheSandDoctor (talk) 04:54, 30 January 2018 (UTC)

That parsing engine is seriously complex. I’m not sure how much luck you will have dissecting it. It’s also deeply integrated into the core.—CYBERPOWER (Message) 05:17, 30 January 2018 (UTC)

The parsing engine core is here.—CYBERPOWER (Message) 05:22, 30 January 2018 (UTC)

Cyberpower678 That was my concern too. Going to figure out a fix somehow to this. Just is going to take some time . I have yet to figure out a way to properly restrain the regex pattern matching. --TheSandDoctor (talk) 05:39, 30 January 2018 (UTC)

I have managed to fix the issue where the hyphen prevented date recognition, but still have an issue where I cannot restrict the regex. If it doesn't have a parameter, it doesn't care the template type, if it has the parameter it will just grab at random in the page (for date parameters). This is frustrating. --TheSandDoctor (talk) 06:03, 30 January 2018 (UTC)

I found a better alternative to incorporating Cyberpower678's parsing engine. While the bot is still reliant on REGEX, it is now somewhat less reliant on them, favouring built in PHP methods (regex to ID the template, PHP methods to figure out its components etc). The way I have implemented it has seemingly (more testing to do) solved both issues raised by GreenC and Keith D. Once further testing is done (looks very promising so far), I will run some limited on Wiki tests on pages that it previously had issues with and update you all one final time. --TheSandDoctor (talk) 05:52, 31 January 2018 (UTC)

If you'e interested how WP:WAYBACKMEDIC parses. Regexing a full article at once is complex and error prone. Break the problem down into steps. Using a scratch copy of the article, delete wikicomments (bad stuff there). Then using regex, extract all citations and delete them from the article so they don't get seen by future steps. Then for each extracted citation, do operations using regex (extract URL, checks for dead-link tags etc). Then extract/delete all URLs surrounded by [] which are outside a citation. Then extract/delete all urls that are left (bare urls). It uses regex at each step, but fairly simple and error-free since it's operating on a restricted domain of text. -- GreenC 15:26, 31 January 2018 (UTC)

@GreenC:@Keith D:@Cyberpower678: I went and translated the bot to Python (source) and have managed to address all the issues raised so far. If You Ever Think I Will Stop Goin' In Ask Double R fix, Steve McManaman double deadlink fixed (refer to last change in that log - ignore the rest; ran 3 times, only one edit - meaning issue resolved), fixed issue (ran bot in between those two edits, it didn't edit - therefore avoiding blanking content) where user error (not closing template) could result in partial article blanking. I hope that this addresses concerns raised and I apologize for the delay in my response, I have been quite busy with uni and have not had the time to rewrite the bot until now. --TheSandDoctor (talk) 04:31, 12 March 2018 (UTC)

From experience the long tail of edge cases is very long due to the informal nature of data on Wikipedia ("anyone can edit") good practice to keep a close eye on the bot edits for new problems cropping up. 3 years of constant work on my bot with over 25k lines of code and still finding new problems.. found a useful common.js script for monitoring a large number of diffs is User:Writ Keeper/Scripts/commonHistory.js -- GreenC 13:10, 12 March 2018 (UTC)

Thanks for the script GreenC! I have just added it. Will definitely come in use (makes inspecting them easier....and adds a big green bolded text diff button....aww, you shouldn't have, green is my favourite colour! ...nice username btw! :D). All joking aside though, thank you for the script, will be useful, and of course I will continue to monitor the bot. At some point in the next little while I will have to generate a new list of articles for the bot to run over, I shall be watching closely (but, as always, I thank people for bringing forward bot issues and diffs, it is honestly a really big help).

To everyone: thank you so much for bringing forward the issues (and diffs) that you have, (not sure how to word this part) I hope that you continue to do so if you notice other errors. --All the best, TheSandDoctor (talk) 18:01, 12 March 2018 (UTC)

NPP Backlog Drive Appreciation

		Thank You
		Thank you for reviewing articles during the 2018 NPP New Year Backlog Drive. Always more to do, but thanks for participating. — Insertcleverphrasehere (or here) 05:46, 31 January 2018 (UTC)