Wikipedia talk:STiki/Archive 19

This is an archive of past discussions on Wikipedia:STiki. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 15

←

Archive 17

→

Vandalism alerts

This is an idea for using STiki queues to alert vandalism patrollers, whether or not they are users of STiki.

At present we have two tools for informing patrollers of the current level of vandalism. Both have been developed by A930913, and both are derived from ClueBot NG.

User:DefconBot: Gives a "Defcon" vandalism level, 1 (highest) to 5 (lowest), based on the current ClueBot Vandal Score (CVS). CVS measures the minute-by-minute sum of the vandalism score (0 to 1) that ClueBot assigns to every edit. In general, Defcon tracks the daily and yearly variations in general traffic.

Am I Needed?: This gives a Yes or No answer to whether CVS score is exceeding "Reverts Per Minute" (RPM), the rate at which ClueBot itself is reverting vandalism at the current time.

What neither of these tools can tell us is whether or not human patrolling is keeping up with the suspected vandalism remaining after ClueBot has done what it can.

It occurs to me that we could measure this by counting the edits waiting to be patrolled at STiki. Could a process be developed to monitor the queue and alert us when a backlog of possible vandalism is building up? In a similar way to Defcon, it could output a signal ranging from "All hands to the pumps" down to "all under control, do something else": Noyster (talk), 19:08, 8 January 2015 (UTC)

@Noyster: Am I Needed? does exactly that. If we assume that ClueBot catches a fixed percentage of vandalism, then we can compare it to the reverts to determine if human patrolling is catching the rest of it. The tool then displays this in a simple Yes/No format. Note also that RPM is all Wikipedia reverts and CVS is the sum of the scores of all edits; including those that aren't reverted by ClueBot.

If you want to see some more stats, check out the statistics section on my user page. 930913 {{ping}} 00:31, 9 January 2015 (UTC)

Thanks, if I'm understanding right, Am I Needed? tells us whether the backlog of edits needing patrolling has grown or shrunk during the last few minutes. This is useful to know. What I'm asking is if we could monitor the STiki queue to present, in addition, a live indicator of the total size of said backlog: Noyster (talk), 13:34, 10 January 2015 (UTC)

I think that the tools of User:A930913 approximate the quantities we are interested in. The STiki edit queue is always more than (several) million elements in length. Consider that the largest the STiki queue can be is the number of NS0 titles on English Wikipedia (which is in the millions now?). If someone reverts an edit, that revert itself overwrites the reverted edit in the queue (although with a score so low it is exceedingly likely never to be seen). Only "innocent" classifications reduce the queue size, but only until there is a new edit on the article in question. What we could measure is "the number of edits in the queue with a vandalism probabily > x", but again, I'm not sure that is a terribly different result than what existing tools are doing. The existing DefCon box is pretty pervasive across user pages, do you think our efforts would result in traction? West.andrew.g (talk) 21:23, 10 January 2015 (UTC)

@West.andrew.g: At 12:25 GMT today, what was the highest score in the queue?

While I have the capability to view the queue at any particular time in hindsight, that involves replaying more logs than I care to think about. Instead, I issued the following query just a couple minutes before this posting:

Query code

SELECT


SUM(case when SCORE <= 1.00 AND SCORE > 0.95 THEN 1 ELSE 0 END) as TIER01,
SUM(case when SCORE <= 0.95 AND SCORE > 0.90 THEN 1 ELSE 0 END) as TIER02,
SUM(case when SCORE <= 0.90 AND SCORE > 0.85 THEN 1 ELSE 0 END) as TIER03,
SUM(case when SCORE <= 0.85 AND SCORE > 0.80 THEN 1 ELSE 0 END) as TIER04,
SUM(case when SCORE <= 0.80 AND SCORE > 0.75 THEN 1 ELSE 0 END) as TIER05,
SUM(case when SCORE <= 0.75 AND SCORE > 0.70 THEN 1 ELSE 0 END) as TIER06,
SUM(case when SCORE <= 0.70 AND SCORE > 0.65 THEN 1 ELSE 0 END) as TIER07,
SUM(case when SCORE <= 0.65 AND SCORE > 0.60 THEN 1 ELSE 0 END) as TIER08,
SUM(case when SCORE <= 0.60 AND SCORE > 0.55 THEN 1 ELSE 0 END) as TIER09,
SUM(case when SCORE <= 0.55 AND SCORE > 0.50 THEN 1 ELSE 0 END) as TIER10,
SUM(case when SCORE <= 0.50 AND SCORE > 0.45 THEN 1 ELSE 0 END) as TIER11,
SUM(case when SCORE <= 0.45 AND SCORE > 0.40 THEN 1 ELSE 0 END) as TIER12,
SUM(case when SCORE <= 0.40 AND SCORE > 0.35 THEN 1 ELSE 0 END) as TIER13,
SUM(case when SCORE <= 0.35 AND SCORE > 0.30 THEN 1 ELSE 0 END) as TIER14,
SUM(case when SCORE <= 0.30 AND SCORE > 0.25 THEN 1 ELSE 0 END) as TIER15,
SUM(case when SCORE <= 0.25 AND SCORE > 0.20 THEN 1 ELSE 0 END) as TIER16,
SUM(case when SCORE <= 0.20 AND SCORE > 0.15 THEN 1 ELSE 0 END) as TIER17,
SUM(case when SCORE <= 0.15 AND SCORE > 0.10 THEN 1 ELSE 0 END) as TIER18,
SUM(case when SCORE <= 0.10 AND SCORE > 0.05 THEN 1 ELSE 0 END) as TIER19,
SUM(case when SCORE <= 0.05 AND SCORE > 0.00 THEN 1 ELSE 0 END) as TIER20,
COUNT(*) as QUEUE_SIZE

FROM queue_cbng;

For which the results were:

STiki CBNG queue by vandalism probability score (tier sizes)
Score	0.95-1.00	0.90-0.95	0.85-0.90	0.80-0.85	0.75-0.80	0.70-0.75	0.65-0.70	0.60-0.65	0.55-0.60	0.50-0.55
Number of edits in queue	0	1	1	8	18	10	14	14	13	23

Score	0.45-0.50	0.40-0.45	0.35-0.40	0.30-0.35	0.25-0.30	0.20-0.25	0.15-0.20	0.10-0.15	0.05-0.10	0.00-0.05
Number of edits in queue	45	211	221	274	1303	7072	30890	50026	70294	921266

TOTAL QUEUE SIZE: 2,876,089

Notice also this is only the CBNG queue. It is not surprising the upper levels have extremely few values (the bot takes care of these automatically, unless the revert would result in an edit war in which case it is given to STiki). I don't think we truly understand what this distribution means. It would be nice to know, for example, "given this distribution, what is the expected revert rate over the next 1000 classifications?" Informally, I get reports every 12 hours about the work being done. I have found if relatively few classifications are done in a day (< 1000), the revert rate is usually about 25%. When we get into high numbers of classifications (>3000) it gets as low as 15%. I want to write an CHI/OpenSym paper that looks at these kinds of STiki statistics and their ramifications in depth. West.andrew.g (talk) 17:58, 12 January 2015 (UTC)

@Noyster: I have STiki reports every five minutes in my IRC channel (##930913 ^connect) if that's helpful. 930913 {{ping}} 12:49, 12 January 2015 (UTC)

Thanks, yes this is one more part of the picture. These are the numbers you get through STiki|Rev. queue|Recent usage stats. They show recent activity by STiki patrollers - the number of edits classified and the percentage classified as vandalism: Noyster (talk), 16:59, 14 January 2015 (UTC)

Hi all. For STiki users, I can see some value in an indicator based on "the number of edits in the queue with a vandalism probabily > x". Am I needed? is good, especially for most tools, which deal with very recent vandalism. STiki has a memory so when vandalism rate goes down and Am I needed? says "no" there may still be a big backlog for STiki users to plough through. Yaris678 (talk) 15:10, 12 January 2015 (UTC)

That is very much the way I was seeing it Yaris. The STiki "queue" is now looking to me more like a bottomless well into which all edits are thrown, those most likely to be vandalism floating to the top to be inspected on STiki. So these are interesting numbers, and we need to examine how they vary over time, and what proportion in each tier ends up being reverted, in order to fix the best score at which to draw the line and arrive at a meaningful measure of total "backlog" at any time. Of course we won't ask for any significant R&D effort on this unless regular patrollers think it's worth while, not just for interest, but for practical use in influencing our work patterns: Noyster (talk), 19:27, 12 January 2015 (UTC)

Also, I am more than happy to dump data (anonymized, if user-names get involved) in order to support this research (and I am shopping for a co-author on the aforementioned paper; Yaris and I attempted previously but were tight on cycles). My own availability is just a little limited at the moment. Thanks, West.andrew.g (talk) 19:34, 12 January 2015 (UTC)

The 10,000 most popular Wikipedia articles of 2014

This fringes on shameless self-promotion, but I think is of broad enough interest to post that my report of the 10,000 most popular Wikipedia articles of 2014 is now available (linking to a talk page with the hope of consolidating discussion).

These type of statistics have relevance in the anti-vandalism domain. On the predictive side, we know that highly popular pages are more likely to be vandalized (although the distribution is somewhat bi-modal, given that pages for smaller EDU institutions also get hit). There are also opportunities for impact-driven queue re-organization. For example, an edit with a 50% chance of being vandalism on a popular article might very well deserve inspection before one a very obscure article that has a 99% chance of being vandalism. Thanks, West.andrew.g (talk) 14:57, 21 January 2015 (UTC)

Not sure about this idea Andrew, because more popular pages will, on the whole, have more watchers and so are already more likely to have vandalism reverted promptly. If the prob of inspection is doubly skewed, by altering queue position as well, then the more obscure articles are more at risk of turning into mush. Vandalism that persists attracts more vandalism. OK they're obscure, but there are very many of them and all will be of interest to someone: Noyster (talk), 16:02, 21 January 2015 (UTC)

Thanks, Noyster. My discussion was hypothetical, and I think there would need to be some practical concessions if it were ever to be implemented.

In my dissertation I used something along the lines of an "expected damaged views" metric in order to discuss this, but I don't recall watchlisters being taken into account. One could model this more completely by looking at prior instances of vandalism on a given page (or pages within a similar range of popularity) that were vandalism-reverted but not identified by bots or probability-based editing assistants. The success of anti-vandalism efforts should not be measured in raw reverts, but instead a measurement like "damaged article views prevented".

Sometimes even mass watchlister/patroller behavior isn't good enough; 20-30 seconds of survival on the most popular pages results in significant exposure. STiki is not the ideal tool for this, but we could imagine that CBNG should be more willing to incur false positives (i.e., lower its auto-revert threshold) for extremely popular pages. Thanks, West.andrew.g (talk) 00:16, 22 January 2015 (UTC)

REP_USER score is alway zero

I was using the stiki api on the raw mode, and I noticed that for as many examples as I tried, the REP_USER score was always 0. Is something broken?

I just queried the database behind for the last 100 edits and I estimate that 1-in-10 had a non-zero user reputation score. If someone has never vandalized (or not done so in a very long time, per a time decay function), their user reputation will be zero. The academic paper that was the basis for STiki explains why this is the case. It is intuitive that for the vast majority of edits, the editor will not have a "criminal history" and the value will be zero. The values lie on [0,1], but an extremely non-uniform distribution requires a more relative interpretation. West.andrew.g (talk) 18:05, 1 February 2015 (UTC)

STiki access please?

Could I please gain access to STiki? I love fighting vandalism on Wikipedia and that is one of my favorite tasks to do. I understand if this is not possible for me at this moment in time, but would it be possible? Thanks a lot! ~HackedBotato ^{Chat with me ♽ Contribs} 03:41, 6 February 2015 (UTC)

@HackedBotato: I think you've demonstrated good-faith in your ~150 article namespace edits. What is lacking for me (in your edit history) is a lack of prior vandalism reverts and an understanding of the warning system. Have you considered WP:CVUA? They could bring you up to speed on the anti-vandalism fundamentals/distinctions quite quickly and I'd be willing to expedite the STiki approval with their endorsement. Alternatively, familiarize yourself with WP:STiki, WP:VAND, and related articles before doing some independent patrolling. I think you are very close here, I just need to know you can apply the same good judgement you've applied elsewhere into the vandalism space. Thanks, West.andrew.g (talk) 15:26, 6 February 2015 (UTC)

Thanks, will try! ~HackedBotato ^{Chat with me ♽ Contribs} 23:56, 6 February 2015 (UTC)

I didn't realize that you needed only rollback to use STiki, turns out I now have that permission! STiki is very effective, thanks for this wonderful tool! ~HackedBotato ^{Chat with me ♽ Contribs} 19:23, 18 February 2015 (UTC)

Hi HackedBotato,

Not realising that there is more than one way to get access to STiki is fairly common. Can you think how we can make it more clear? WP:STiki#Using STiki currently says "The account must have either: (1) the rollback permission/right, (2) at least 1000 article edits (in the article namespace, not talk/user pages), or (3) special permission via the talk page." Do you think a re-word would help.

Yaris678 (talk) 08:22, 19 February 2015 (UTC)

When I first wanted to use STiki, I didn't read carefully and supposed you needed over 1000 edits and rollback permissions, it was just my hastiness. ~HackedBotato ^{Chat with me ♽ Contribs} 19:11, 19 February 2015 (UTC)

User custom messages

I'm sure I'm missing something obvious, but when using STiki, how do you save a custom message? I've tried entering text into "CUSTOM: User custom #1" (after pressing "Good Faith Revert") and it submits that message for that particular incident, but does not save the text for future edits. I can't see a "Default" button or anything to save the text; is there somewhere else you are supposed to insert custom messages? — Bilorv_(talk)^(c)(e) 19:40, 11 February 2015 (UTC)

@Bilorv: The system is designed such that when you enter a message in one of the custom slots, that message is saved for future use. Clearly that isn't happening here, and this is the first issue I'm aware of with this functionality (so I am inclined to believe there is an anamoly in your environment). What version of STiki are you using? On what OS? Do you own the machine on which you are running STiki, or is it institutional hardware? For persistent storage, the messages (and other STiki settings) are written to a (sometimes "hidden") settings file that should exist adjacent to the STiki JAR. I could imagine a situation where you might not have write permissions on that file-system? If you change other settings (text size, checkboxes, etc.), do the changes persist between STiki restarts? Thanks, West.andrew.g (talk) 19:58, 11 February 2015 (UTC)

I'm using version 2.1, having been forced to update a couple of days ago. I'm using Windows 8.1 and am slightly unsure of what "institutional hardware" is supposed to mean, but yes, I own the laptop I'm using to type this and am an administrator on it, so should have permission to change anything. Changing the font size on STiki persists after logging out, exiting and reopening the JAR file and logging in again. I have just tested the custom notice again, and it hasn't saved the message I entered. I copied and pasted the message, if that's relevant. — Bilorv_(talk)^(c)(e) 20:14, 11 February 2015 (UTC)

@Bilorv: Just to satisfy my personal curiosity, type a character in the box (even if you type it and then erase it) after your copy-paste action. That dialogue is configured to save the messages every time there is a keystroke in the box (which copy-pasting with a mouse or CTRL C+V might be getting around?). I have a feeling that isn't going to fix our problem, though. See if you can find a file named ".STiki.props.xml" on your file system. Note that because the filename begins with a period, it is "hidden", and may not reveal itself without some advanced searching options. It is likely to reside: (1) in your "home" directory, or (2) immediately adjacent to where the STiki JAR lives (maybe a talk-page watcher with Windows 8 could confirm the location?). If we can find this file (which is likely, given font size persistence), you should be able to open it in a text editor. Find the variables assigned to the custom messages, what are they? (you should also be able to drop your copy-paste text in directly here). Thanks, West.andrew.g (talk) 16:25, 13 February 2015 (UTC)

@West.andrew.g: When I originally first used the custom message feature, I copied another template (I think "General revert notify") but altered several bits of text, so I don't think making keystrokes will change anything. I can test this again if I you want me to.

I've found a ".STiki.props" (no ".hml") but I'm not entirely sure what I'm looking for. Is <comment>STiki Settings</comment> or <entry key="agf_custom1"/> relevant? Also, it's opened itself in Internet Explorer and through the search method I used, I'm not able to open it in any other format — would this be fine for making changes? — Bilorv_(talk)^(c)(e) 17:41, 13 February 2015 (UTC)

@Bilorv: Yes, that is the file in question. Can you browse to the location using something like "Windows Explorer" and right-click on the file to use the "Open with..." menu option in order to get it open in something besides Internet Explorer? IE is fine for reading, but you'll need to use something like Notepad for editing. We are interested in the keys like "agf_custom1", "agw_custom2", etc. Do these keys currently have values? I am currently on travel using a work laptop, so I don't have access to my usual complement of STiki files. Can a talk-page watcher copy-paste this portion of their settings file for reference? Thanks, West.andrew.g (talk) 16:47, 14 February 2015 (UTC)

I opened the file in Notepad and the "agf_custom[x]" keys didn't have any values. I assumed the next step was to change the value, save and see if it opens in STiki. I changed <entry key="agf_custom1"/> to <entry key="agf_custom1">test</entry>, saved the file and opened STiki. The "CUSTOM: User custom #1" message, contained test. So it works — I've pasted in the custom message I wanted, and it looks like that's going to stay there after multiple uses (although I haven't used it yet). It still doesn't explain why the feature didn't work automatically, but thanks for your help. I suppose the next step would to be ask other Windows 8.1 users whether the feature works for them [again, cue summoning of talk-page watchers]. — Bilorv_(talk)^(c)(e) 18:27, 14 February 2015 (UTC)

Glad to hear everything worked out here, albeit via a circumvented method. I will try to subsequently troubleshoot this, but at least I can give it lower priority in the meantime assuming no one else is experiencing this system behavior (right?). It is very odd this fails for you while other persistent settings still work; I imagine this will be a tricky issue to hunt down. Thank you for your use of the tool. West.andrew.g (talk) 15:39, 15 February 2015 (UTC)

Hmm, not quite working on Mac either. My presets from about 6 months ago are still there, but new changes to the custom messages don't work. Where are the persistent settings on Mac found? --L235 (t / c / ping in reply) 05:07, 16 February 2015 (UTC)

Just in the process of switching to MAC OSX myself. Can a talkpage stalker comment? The code specifies the file should go in the "home" directory, and failing that, should be placed directly adjacent to the executable JAR. Why not just use the spotlight/search to look for the file that Bilorv mentions above, as that is consistent across systems. Given the fact multiple people are experiencing this, I will give the bug a more prioritized look. Thanks, West.andrew.g (talk) 14:04, 19 February 2015 (UTC)

Done - Bug identified, labeled as T#051 in the table above, and then resolved in source. The issue arose when we started pre-pending organizational labels ("TOOMUCH:", "CUSTOM:", "NONSENSE:") to the AGF message title a few versions back. The save/import functionality was still using text without the prefixes. Thanks, West.andrew.g (talk) 01:03, 20 February 2015 (UTC)

STiki access please?

Hello Wikipedians! While I haven't been an editor for a very long time, I have made about ~200 edits in the article namespace. I use Twinkle, and almost exclusively fight vandalism. I am currently enrolled in the CVU academy, under the user Shaun. However, he is semi-retired so there may be question as to how useful he may be as a trainer. I have gotten fairly adept at profiling edits that may be vandalism on the Recent Changes list, and would like this permission to help fight vandalism on Wikipedia! Hoo-Rah. k_scheik (talk) 04:44, 22 February 2015 (UTC)

Done -- @K scheik: -- This user was pretty accurate in their self assessment. Talk-page/contributions revealed an associated ANI report (I searched it out; it was a innocent issue), with the only real negative being a testy exchange with another editor. Given CVUA enrollment and a good bit of experience, I have approved this request. Thank you and happy reverting. West.andrew.g (talk) 20:21, 22 February 2015 (UTC)

Bug Report

Hey, I had my account usurped a while ago and I think it may have caused something in Stiki's system, because when I tried to log into my account on the system, it resulted in my IP being able to use Stiki WITHOUT ROLLBACK OR 1000 EDITS, JUST BY OPENING THE SOWFTWARE. Shaun ♣^༆ 23:45, 22 February 2015 (UTC)

@Shaun: Are you using the most recent version of STiki? We resolved a related bug one or two editions back. I don't fully understand how the usurpation process works from a technical perspective, but that seems unlikely to be the culprit. I also made a minor database tweak regarding your account, so give it a retry if you could. Thanks, West.andrew.g (talk) 17:44, 23 February 2015 (UTC)

STiki access consideration

I've been doing mainly just vandalism control the old fashion way by patrolling recent changes, and I believe I've proven myself in this and the user warning template. I will admit to one small mistake on my part (discussion on my talk page), but besides that I believe I've been pretty reliable and helpful when it comes to fighting vandals. Currently I'm using twinkle and this has been very helpful, but I think STiki could extremely increase my productivity in fighting vandalism. I'm trying very hard to enroll in the CVUA, contacting several trainers currently to no success, but I will continue to try. Regardless of your decision any feedback would be great! Thanks for consideration! Kharkiv07 (talk) 05:11, 24 February 2015 (UTC)

Done -- @Kharkiv07: Access approved. In just three of four days of account history the work has been fast and furious, but the talk page alludes to prior IP experience, with the level of work and syntax knowledge suggesting this is the case. Lots of anti-vandalism and patrol work with only one minor complaint that was handled appropriately. This meets the threshold for me (FYI, the 'edit count' tool is down). Thank you and happy reverting, West.andrew.g (talk) 16:32, 24 February 2015 (UTC)

Articles with deletion template

Hi, I thought this would be rather a relevant thing to bring up. Is it possible to put up an indicator at the top of a Window if the revision that Stiki is showing had a CSD/PROD/AfD tag on the page? Especially CSD, as it would would help people review an under that has been tagged? --Rsrikanth05 (talk) 20:43, 24 February 2015 (UTC)

What do you mean by, "review an under that has been tagged"? How would the presence of such a tag affect vandalism assessments? I'm always willing to help out STiki users, but I am also cognizant of mission creep. Thanks, West.andrew.g (talk) 14:49, 25 February 2015 (UTC)

For instance, if I see an article that has been tagged under CSD/PROD, and it seems certain that the article is going to be deleted, it would help the Stiki user if he/she knows that the article has been tagged as such. For instance, I see some tall claims being made and I don't bother reverting them because the article has been CSD'd as unambiguous advertising/spam. --Rsrikanth05 (talk) 13:33, 27 February 2015 (UTC)

I'm inclined to add this to the tracking table with a medium priority. Thoughts from other users about the idea? Or how it should be represented? I imagine the same purple text area that is used to indicate multiple edit rollback might be a reasonable place? Thanks, West.andrew.g (talk) 05:19, 28 February 2015 (UTC)

I have occasionally come across this issue too. I think that putting text where the purple multi-edit info goes is a good idea. Maybe make the text a different colour (green?) so that people don't just think that it is the same old message.

Yaris678 (talk) 15:04, 28 February 2015 (UTC)

I have also had the occasional issue regarding tagged articles (especially CSD tagged articles). Huggle, for example, always lets you know when you're about to revert an edit which was made by the same user who created the page in question. This is especially useful when it comes to page blanking. Knowing if an article is a CSD candidate, or a new article created by the same user in the diff, can have a significant outcome on the decision made, when using anti-vandal tools. Orphan Wiki 20:33, 28 February 2015 (UTC)

I really like this idea- I come across this all too often and often spend quite some time Googling it until I actually look at the page and it happens to be speedily deleted by then. Cheers, --L235 (t / c / ping in reply) 23:39, 28 February 2015 (UTC)

Tracked in table as T#052. Thanks, West.andrew.g (talk) 01:17, 2 March 2015 (UTC)

Stiki down?

Is Stiki down? I am able to login, but I am not bring present =ed with any diffs. --Rsrikanth05 (talk) 06:45, 1 March 2015 (UTC)

Never mind. It's working now. --Rsrikanth05 (talk) 11:00, 1 March 2015 (UTC)

The STiki database is backed up early Saturday mornings (EST time). A lock is placed on several critical tables in order to keep the copies consistent. This normally lasts only a few minutes, but you may not get served a diff while this is ongoing. West.andrew.g (talk) 01:07, 2 March 2015 (UTC)

@West.andrew.g: How easy is it to run a STiki backend server? That is, does it take a lot of resources? If not, I could donate some computing power or run a backup backend in case your hosting failed for some reason. --L235 (t / c / ping in reply) 05:26, 2 March 2015 (UTC)

Thanks for the offer, but the infrastructure is quite stable. STiki is run out of the University of Pennsylvania, and when we do run into problems, it is not about what UPenn provides but the fact I often need to have a colleague physically access the machine (an issue we are trying to eliminate with some troubleshooting). STiki's data is backed up to a couple of places distributed around the globe. If it should be migrated anywhere it should be the WMF Labs machines, but I don't have the time or energy for that at the moment. Thanks, West.andrew.g (talk) 15:11, 2 March 2015 (UTC)

STiki not taking in edits?

I am seeing diffs as of now (9 March 2015), but they are all from over 6 (and up to 30) days ago. Normally, they are within the last hour. Is there something going on with STiki servers? Iwilsonp (talk) 01:22, 10 March 2015 (UTC)

That's been happening for a few days now. It happens quite frequently actually. Widr (talk) 07:45, 10 March 2015 (UTC)

Done - I can confirm that edit ingestion was stopped for a little while. I restarted it several hours ago and things should be back to normal. I think STiki's backend service boots up with 32-64 threads assigned to the work of edit ingestion. Occasionally one of these threads "stalls" (it isn't an infinite loop because I track CPU usage, and it isn't failing altogether, as I have exception handling and thread-restart for that) and is rendered ineffective. Every 2-4 weeks enough threads have stalled such that an insufficient number remain inactive to keep up with the editing pace and we start to get behind. I look it the queue size 4-times daily via emailed reports to track this behavior and I normally restart the service when things start to appear symptomatic. In the case of this weekend, I was in a position where I was unable to read or act on those emails (which were showing the issues). I realize the inelegance here, but tracking an anamoly that might get triggered by only a single edit each day has proven difficult. I suppose I could also write a script that forces a hard restart of the STiki service once daily/weekly. Thanks, West.andrew.g (talk) 05:31, 12 March 2015 (UTC)

I think STiki is being firewalled by my org!

Everytime I try to connect, I get an error about not being able to connect to the STiki servers. I can do it at home, however. Just my two cents, that it DOES get blocked by some orgs, so ticket 4 is an issue. Grognard Chess (talk) Ping when replying 05:25, 27 March 2015 (UTC)

Acknowledged, but it also requires a good deal of work on my side to make the transition and will lead to some performance degradation (i.e., we will still default to a persistent connection with the SQL server). You could always seek out a sympathetic sys admin to make a firewall exception, assuming STiki is the type of thing the organization would endorse on their time. There are ways you can evade these blocks using port tunneling and a proxy server, but realize that takes some technical know-how. West.andrew.g (talk) 15:43, 27 March 2015 (UTC)

Request for access

(retitled by me -Ugog Nizdast (talk) 08:34, 28 March 2015 (UTC))

Hello. I was just going through some articles that I found several clear vandalism signs particularly by IP Addresses. I can't revert these edits simply by Wikipedia's undo tool as the vandal has made several edits into the article. This is just an instance, I have been through many such vandalism cases where I cannot do anything, but sit hands-folded, waiting for someone to revert it. And by the time someone does revert it, the wrong information is exposed to numerous readers of Wikipedia. I believe 70% of the reason why Wikipedia is said to have wrong information is because of vandalism which is not treated. For another reason, I want to treat vandalism. I want to help curbing this menace. This tool will help me a lot and do the work easier. I will be able to handle lot many cases in a short time. Please provide me with the access to this tool as I don't have 1000 edits on Wikipedia, BUT I am not a beginner, I can surely handle this tool as my overall edits cross 1000, but the edits in article namespace don't. It will be very generous for you to do the same.--Kashish Arora (talk) 03:26, 27 March 2015 (UTC)

@Kashish Arora: Actually, there are plenty of ways to revert multiple edits at once. For example, you can click on the diff you'd like to revert back to, click "edit", and then immediately save. This'll get rid of all intermediary edits. You can also install Twinkle, a gadget that'll let you click one button to revert to a previous diff. Grognard Chess (talk) Ping when replying 05:20, 27 March 2015 (UTC)

@Chess: Alright, I get it. Thanks! But, it was just one reason, I may edit my request.--Kashish Arora (talk) 05:46, 27 March 2015 (UTC)

I'm a bit confused, is this a feature request? or an access request? The justification here isn't the most coherent and the user in question seems to have little anti-damage experience (as demonstrated by the initial question). That being said, the user seems to be very close to 1000 article namespace contributions, (aside: Why isn't the edit count tool working? and if it is broken, why is it still linked on "contributions" pages?) so maybe the permission should be extended as a sign of good faith? West.andrew.g (talk) 20:11, 27 March 2015 (UTC)

I think what User:Kashish Arora wants is access. I checked the past month's user contribs randomly and I don't think there's enough experience in anti-vandalism activities here. By the edit count, I think the 1000th is about to be/already crossed, so I don't think special permission is needed.

Kashish Arora, although I see that you're well experienced in regular editing, I still think you should demonstrate your counter-vandalism skills. For that, WP:VANDALISM should be read; you should find, revert and warn (and of course, report) where you find such instances. No doubt, you'll have access to STiki soon, but I personally recommend you get first some experience in manually finding vandalism or else you might end up making mistakes using it. Ugog Nizdast (talk) 08:34, 28 March 2015 (UTC)

I have as such not reported vandalism, but I have corrected it and informed users personally about it, by typing the whole of a message to them. This experience has given me a chance to curb vandalism furthermore. And, about the edit counter, I don't really see it working alright. I moved my userpage from Kashisharora11 to Kashish Arora recently. Since then, I am surely facing some little problems regarding some scripts which have edit counters. But, I am consistently trying to protect pages. I have virtually protected the article Chandigarh myself including other articles mentioned on my talk page. I have handled all instances including removal of content, replacement with incorrect material, vandalizing by purpose and so on. I shall and may do great with this tool. But, I am not able to access it, even if I have reached 1000 edits which I think I should have reached by now. I couldn't wait anymore for getting the tool activated for me even after my reaching 1000 edits. So, I have requested this feature.--Kashish Arora (talk) 08:55, 28 March 2015 (UTC)

Have you downloaded STiki and tried to login? You have 1,042 mainspace edits, so you should be able to use it now. Melonkelon (talk) 09:24, 28 March 2015 (UTC)

I have downloaded STiki and I don't know why but I think I am getting the Unix one with the file as "java -jar STiki_exec_[date].jar", but that too isn't .JAR, it is a .RAR file. This .rar file contains all the code, but unless I have an executable application, this code is of no use.--Kashish Arora (talk) 12:48, 28 March 2015 (UTC)

I not quite sure I get what you mean. I'm assuming you downloaded the .jar file right? And you're saying that it's showing as a .rar file? If I'm not wrong, on Windows it's directly executable, while on Linux it can executed in the command line using the command. Andrew? -Ugog Nizdast (talk) 10:51, 29 March 2015 (UTC)

Wait, I think I know. On Windows, STiki should be executable normally. For some reason, you see it as a .rar file. Have you tried the "Open with" option? There should be some way to execute it using Java. -Ugog Nizdast (talk) 14:02, 29 March 2015 (UTC)

@Ugog Nizdast:, I have tried that for sure. I have tried opening it using java.exe, javaw.exe, javaws.exe. The only error it shows during the process of opening with these applications is that the file is basically .RAR.--Kashish Arora (talk) 12:52, 30 March 2015 (UTC)

The downloaded file is a .ZIP. You need to unzip that to get to the JAR file. West.andrew.g (talk) 19:43, 30 March 2015 (UTC)

@West.andrew.g: I know I have to unzip. But, after unzipping, there is another .RAR file inside the first one where everyone is saying it is supposed to be .JAR. I have downloaded a dozen times, zipped the first file just to find there are two .TXT files (CHANGELOG and README) and there is another .RAR file which contains all the code. I am somewhat good at computers and I surely know the file is .RAR and not .JAR.--Kashish Arora (talk) 04:32, 31 March 2015 (UTC)

There is no .rar file in the .zip file. There's a single .jar file (STiki_2014_12_12.jar) and two .txt files. Melonkelon (talk) 04:20, 1 April 2015 (UTC)

@Melonkelon: Well, there is one. I can even upload a snip if you want.--Kashish Arora (talk) 05:55, 1 April 2015 (UTC)

There is not a RAR file in a legitimate copy of the download. As Melonkelon stated, there are two TXT files and one JAR. A JAR or "Java Archive" file is fundamentally similar to a RAR in that it is an abstraction for groups of files. Thus, it wouldn't surprise me if a RAR unpacking utility tried to own the JAR extension and offered to unpack the file (which it can do, but this is not the same as the 'execution' we desire). This would happen if: (1) A JRE (Java Runtime Environment) has not been installed on the machine; i.e., you need to have Java -OR- (2) Even if Java is installed, a (UN)RAR utility could be the primary association of the JAR extension. If this is the case, a right-click and "run with..." option might reveal the option to execute the file using Java. Not sure the OS we are dealing with here, but command-line execution is always an option. West.andrew.g (talk) 21:13, 1 April 2015 (UTC)

@Kashish Arora: Please attach screenshots if you are continuing to have difficulty. Have you attempted command-line execution? Thanks, West.andrew.g (talk) 18:14, 3 April 2015 (UTC)

@West.andrew.g: Done! Finally! I converted .RAR file to .JAR file with one of the online tools. I can finally have access to STiki! Thank you all.--Kashish Arora (talk) 04:50, 4 April 2015 (UTC)

Request for permission

While as of the writing of this my total main space edit count numbers only 139, almost all of those are edits relating to anti-vandalism work. In addition to this, I was briefly enrolled in the CVUA however I was unable to complete it due to the departure of my mentor, I too went inactive around the same time though I have since returned. I believe As such I now feel I have amassed what I believe to be sufficient anti-vandal experience, though please correct me if I'm wrong. I request access to STiki because I believe that it would enable me to more effectively combat vandalism and accomplish more in a shorter amount of time. Roborule (talk) 01:10, 5 April 2015 (UTC)

Done -- Nearly 750 edits due to work in "Talk" and admin spaces. Enrollment in CVUA demonstrates right attitude. Virtually all contributions are for patrol tasks. Happy reverting. West.andrew.g (talk) 01:37, 5 April 2015 (UTC)

@Roborule: Forgot to ping in initial approval. West.andrew.g (talk) 01:39, 5 April 2015 (UTC)

Pronunciation

How do we pronounce STiki? Every time I use it, read its name, or want to speak it, I hesitate. Is it Ess-Ticky or Sticky or anything else?--Kashish Arora (talk) 09:26, 4 April 2015 (UTC)

I've always said it like "sticky". Yaris678 (talk) 16:50, 4 April 2015 (UTC)

Yep, that's how I've always said it, and this is probably on the video record of some prior Wikimania presentations. That's as close to official as we are going to get. Thanks, West.andrew.g (talk) 17:17, 4 April 2015 (UTC)

I read it as "ess-TEE-kee". >_> — Jeraphine Gryphon ^(talk) 13:24, 20 April 2015 (UTC)

Request for STiki

Even though I have few namespace edits (about 220 the last time I went on user analysis), almost all my edits are anti-vandal work. It is hard and tedious using the Recent changes without anti-vandalism tools (like Huggle), and quite slow (almost half of the users I catch vandalizing have been reverted by someone else with Huggle). I already use Twinkle,but I don't think it is enough. I would like to use STiki to help me and speed up my anti-vandalism work. May I use the STiki tool?--ABCDEFAD^{talk to me} 21:37, 24 April 2015 (UTC)

Real life is busy right now, can an experienced STiki user and talk-page stalker give me a recommendation in keeping with my usual assessments? Thanks, West.andrew.g (talk) 21:08, 27 April 2015 (UTC)

Stalked the user. Request for rollback was denied a week ago but the user has displayed continuous anti-vandal work (warning and reporting) in the past few days, and spot checks didn't find any blatant mistakes. This looks fine to me and I support the request. -Ugog Nizdast (talk) 09:28, 28 April 2015 (UTC)

He has used a number of creative templates and also used twinkle. He should be allowed to use Stiki. OccultZone (Talk • Contributions • Log) 09:35, 28 April 2015 (UTC)

Has only been active in article space for 15 days but has made 298 reverts, and 14 reports to AIV all of which resulted in blocks. I'd endorse this, with cautions: talk page gives evidence of two "false positives" in recent days, and STiki makes it very easy to multiply these if not used judiciously. The editor is urged to (re-)read all the words of wisdom under WP:VANDALISM and consider branching out into other types of editing, to broaden experience of Wikipedia: Noyster (talk), 09:58, 28 April 2015 (UTC)

Done -- @ABCDEFAD: Access approved on basis of community support. Thank you everyone for your assessments. Thanks, West.andrew.g (talk) 16:28, 28 April 2015 (UTC)

Speedy deletion

I don't actually know if it's a bug or a feature or anything that requires attention, but lately I've noticed a few times that STiki freezes completely whenever you come across an article heading for (speedy) deletion. Whenever I come across the deletion tag, I usually study the page via browser, and if the article is deleted by the time I return to STiki, I have to close STiki and start it again in order to get new diffs. Probably nothing that needs action, but interesting observation anyway, so I decided to share it with you fellow STiki users. Widr (talk) 17:30, 30 April 2015 (UTC)

How do you know it is a CSD candidate in the STiki window? The articles are so short you are seeing the markup for the tag and the edit? How often does this happen? I'm a bit surprised that young articles with little/no reputation history are making it atop the queue. Could you save the RIDs the next few times you encounter this? The freeze happens after you perform a classification action? Or just when you return to the window? Does the classification action chosen effect the bug? If the article isn't deleted, then everything goes through okay? West.andrew.g (talk) 18:03, 30 April 2015 (UTC)

I know it, because there is the "db" tag at the top of the article. This only happens if I don't make any actions with STiki. When I return to STiki and the page has been deleted in the meantime, that's when the program freezes and won't move on to the next article. Widr (talk) 18:24, 30 April 2015 (UTC)

AIV problems

I have been using STiki, and I recently reported an IP editor to AIV. Then, the editor vandalized again and I reverted him/her. Then, I check the AIV page and find that I reported him/her twice. In your next STiki release, can you fix this problem? --ABCDEFAD_{talk to me} 17:58, 29 April 2015 (UTC)

In my opinion, this is somewhat of a non-issue. Notice the presence of User:HBC_AIV_helperbot through the WP:AIV page history. One of that bot's primary tasks is combining duplicate reports. You simply "beat" the bot to its next pass of analysis (which seems to happen very frequently). Could I independently write code to obtain and parse the AIV page for all reported users/IPs before STiki edits that page? Sure. But why put in that effort when there is already a bot running that doesn't require cycles on the client-side or programmatic effort? Moreover, I'm not sure other tools make these checks, meaning the same problem can arise via multiple users/channels. Thank you for your feedback, but given the existence of that bot, I do not consider this high priority. Thanks, West.andrew.g (talk) 18:26, 29 April 2015 (UTC)

Also, there aren't many templates; may you please add more?--ABCDEFAD_{talk to me} 18:17, 29 April 2015 (UTC)

AGF templates? These are agreed upon in a community manner and suggestions should be discussed on this talk page. The main WP:STiki page links to a sub-page where the AGF messages are all written out and collaborated upon. West.andrew.g (talk) 18:28, 29 April 2015 (UTC)

Is there a 4im hotkey? Sorry to bother you, but I didn't see it in the article.--ABCDEFAD_{talk to me} 21:34, 30 April 2015 (UTC)

4 and b map to "4im" and "back", respectively. Thanks, West.andrew.g (talk) 15:30, 1 May 2015 (UTC)

Renamed

I was renamed from "Lixxx235" to "L235" a while ago, so could my classifications from the two usernames be merged into the "L235" username? Thanks. --L235 (t / c / ping in reply) 11:55, 1 May 2015 (UTC)

Done - Remapped 599 classifications. Changes will be reflected when the leaderboard is regenerated in ~12 hours. Thanks, West.andrew.g (talk) 15:40, 1 May 2015 (UTC)

STiki hangs when logging in, can't login on slow connections

Pretty much what the headline says: STiki always hangs when I'm logging in, even on a good connection, and it's pretty much impossible to login when it's slow. It used to be fine, but somewhere down the line, it just started hanging and not doing anything. It also hangs when I classify the first edit that is displayed, though it's only brief, like 3 seconds. I don't think it's just my computer or network because I've also tried on a different computer at uni, and same problem. Does anyone else experience this? Melonkelon (talk) 21:08, 28 April 2015 (UTC)

Hello, from here STiki works fine, no slow downs or anything. So yours is not a problem in general. Take care.Super48paul (talk) 07:32, 29 April 2015 (UTC)

Happens with me too! Though I have a 4GB RAM system (which obviously isn't required by STiki), the problem still persists. The 'hang' is observed during logging in and while first classification. We can't even choose 'keep me logged in' for STiki.--Kashish Arora (talk) 13:49, 29 April 2015 (UTC)

Hmm, I am not surprised to hear about a "slight delay" of a couple seconds upon login and the first classification -- I am surprised to hear that some people experience a significant enough delay that it is a serious usability issue. When someone logs in, there is a small handful of queries that need to be made: the two-phase login via Mediawiki, edit count, explicit block/permission checks, confirmation of native rollback capability, and maybe one or two other small things. When the first classification is made after logging in, more needs to be done: the entire edit cache and queue reservation is cleared because the now logged-in user may hold a "pass/ignore" status on the currently cached edits, a new queue reservation block is obtained respecting prior "ignore" annotations, and the client probably starts ~10 threads simultaneously with the work of obtaining those edits data and processing/caching them for GUI display. Usually this work is ongoing in the background to keep the cache full, but the latency becomes apparent when a bunch of it is being done in parallel up front.

How much of a delay are we talking about in the worst case? While I see the benefit of a "stay logged in" type option, this isn't the big latency sink, and that puts a tremendous burden on me to do credential storage in a secure fashion. How many times a day are folks opening and closing STiki that this becomes an issue? Thanks, West.andrew.g (talk) 15:21, 29 April 2015 (UTC)

I clocked the delays involved. Opening STiki: 3 seconds; login: 3 seconds; first edit classification acceptance: 15 seconds; subsequent acceptances: less than a second. So only the first acceptance is noticeable. But we patrollers accept that gladly, because after that it is all smooth sailing. So a non-issue indeed.Super48paul (talk) 07:47, 30 April 2015 (UTC)

@West.andrew.g: I am a programmer too, a beginner maybe, but I do understand how much processes STiki would be undergoing before opening up. But, we can surely improve it. So here are the timings which STiki took for me: Opening- 8 seconds, logging in- 5 seconds, first classification- 15 seconds. That first classification really lags a lot.--Kashish Arora (talk) 12:40, 30 April 2015 (UTC)

My thinking is that 99% of these costs aren't computational, but network latency. Unlike the vast majority of services one tends to use on the Internet, STiki is a single machine, in a single place, that is not served out of a CDN. I'm willing to bet that those who experience long delays are far away from Philadelphia, PA, USA where STiki lives. I can't make the Internet quicker, and the program is already quite well multi-threaded. There is no simple way to fix this aside from reducing the number of queries, which is something that has to be constantly balanced against incoming feature requests. West.andrew.g (talk) 16:23, 30 April 2015 (UTC)

Living in Australia doesn't seem to help. Logging in used to be very quick for me, and I could sign in every time, even on a slow connection, but subsequent changes to STiki seem to have affected it to the point I can't login at all. It sounds like it won't be an easy fix, so I guess I'll just have to use it when my internet is fast. Melonkelon (talk) 01:16, 1 May 2015 (UTC)

I don't think I've independently made changes to the login procedure in... years? I've been forced to change the code (and push emergency releases) because of sudden API changes by the WMF (including a move from a one query process to a token-based one that requires two calls). However, these are communications between the client and the WMF servers exclusively, and should be quicker than a round-trip to Philadelphia. I'll delve into the code later to rethink everything that goes on when that button is pressed, but I currently can't explain a historical slowdown. Thanks, West.andrew.g (talk) 15:55, 1 May 2015 (UTC)

Default web browser

How does it handle this? My default is set to Chrome (I've switched from Firefox recently) but it still uses Firefox when I click on any of the links for editing outside Stiki. Ugog Nizdast (talk) 02:27, 6 May 2015 (UTC)

It is non-trivial. The code STiki uses is here. Basically there is a package called Java.Desktop that should handle things in most instances (how it determines which browser to use, I don't know). Absent that, STiki will try some MAC OSX and Windows specific calls to get at the default file handlers. Absent that, it just starts seeing which browsers might be present on the machine. If you're on a modern machine and Java distribution you likely fall into the first bin. Quick Internet searching suggests there might be a setting in the "Java control panel" to set the path to the preferred browser. West.andrew.g (talk) 15:27, 6 May 2015 (UTC)

Done it at last, set back everything to Chrome. Did some snooping around with the keywords "java desktop setting default browser". -Ugog Nizdast (talk) 17:20, 6 May 2015 (UTC)

Maintenance?!

STiki is down from here. Any maintenance going on? I shall just rest. Huggle works all right, so W-servers are functioning. Super48paul (talk) 08:02, 2 May 2015 (UTC)

My STiki is down too! --JAaron95 | Talk | Contribs 12:14, 2 May 2015 (UTC)

I think it's the periodic maintenance again. --Rsrikanth05 (talk) 13:34, 2 May 2015 (UTC)

Confirmed down, but for what? --I am k6ka ^{Talk to me!} _{See what I have done} 18:29, 2 May 2015 (UTC)

STiki is hosted on the UPenn servers, which is down for maintenance once in a while. --Rsrikanth05 (talk) 11:07, 3 May 2015 (UTC)

It is early in the morning at UPenn. Still too early to wake up the servers?! Still no STiki sign of life.Super48paul (talk) 14:11, 4 May 2015 (UTC)

Done -- STiki has life. More description to follow when I get everything stabilized. West.andrew.g (talk) 15:54, 4 May 2015 (UTC)

I can't provide too much clarity into the situation other than to say everything should be back up and fully functioning. Around 11PM EST on MAR-1 I received notification the server network connection had been lost. This situation affected several of my colleague's servers on the same network. To the best of our knowledge no maintenance window was announced, nor has an explanation been offered since. Power was not lost to the machine -- it knew it was offline -- and made unsuccessful attempts via the NetworkManager to re-attach itself. Needless to say, a Friday night was the worst time this could have happened. A human peer intervened Monday morning and everything seems to have come back online just fine. Sorry for any inconvenience; I will continue to experiment with the self-healing script. West.andrew.g (talk) 14:56, 5 May 2015 (UTC)

Andrew, you've done a tremendous job out here! Thanks a lot for all the hard work. :) --Rsrikanth05 (talk) 16:49, 7 May 2015 (UTC)

Requesting special access to STiki for a Wikipedian I am teaching

Hello, I'm currently teaching a fellow Wikipedian, User:DweltGraph, about vandalism and how to handle it. I fully trust him, and I will take full responsibility for anything that goes wrong. I have already gone through everything he needs to know about vandalism on Wikipedia, and he has practiced using WP:Twinkle with me (in-person). I think STiki is a great tool, and that's why I think it's essential to teach it to him. I hope you consider this, and don't hesitate to not accept this if you don't find it fit. Thanks! ~HackedBotato ^{Chat with me ♽ Contribs} 14:17, 19 May 2015 (UTC)

I presume that a few hundred edits more and he will be a fine patroller... What is the hurry?!Super48paul (talk) 14:25, 19 May 2015 (UTC)

@HackedBotato: Where in his edit history is this Twinkle experience you speak of? West.andrew.g (talk) 20:36, 20 May 2015 (UTC)

@West.andrew.g: I taught him in-person and he learned how to use Twinkle well. I'll get him to start reverting vandalism on his account. ~HackedBotato ^{Chat with me ♽ Contribs} 20:45, 20 May 2015 (UTC)

Access

I have downloaded STiki and installed java, can I please have special approval to use it. Thanks TeaLover1996 (talk) 05:02, 27 May 2015 (UTC)

Hi TeaLover. You have 756 mainspace edits, which is most of the way to getting STiki access automatically. I am inclined to say that you should be given access now. The only thing that is making me cautious is that quite a lot of your edits don't have any edit summaries. Do you have anything to say about that? Yaris678 (talk) 12:51, 27 May 2015 (UTC)

@Yaris678: When I use pending changes and I'm reverting an edit, it says if the edit is not vandalism an edit summary should be used so usually I will do that, if it is vandalism I just revert the edit and sometimes use an edit summary to display a revert of vandalsim. TeaLover1996 (talk) 14:45, 27 May 2015 (UTC)

I agree that reverting obvious vandalism doesn't need additional information. I am more concerned with edits like this. Other editors might appreciate some kind of explanation of why you have removed the wikilink. I'm not saying there was anything wrong with removing the wikilink, just that an edit summary would be helpful. Yaris678 (talk) 15:36, 27 May 2015 (UTC)

@Yaris678: Will do. Any decision or are you leaving it to someone else. Regards? TeaLover1996 (talk) 15:48, 27 May 2015 (UTC)

Done -- @TeaLover1996: -- You now have access to the tool. West.andrew.g (talk) 16:35, 27 May 2015 (UTC)

Access (also)

I'm requesting STiki access for my alt account Jeraphine (talk · contribs · deleted contribs · logs · filter log · block user · block log), so I can use STiki from there. — Jeraphine Gryphon ^(talk) 09:58, 27 May 2015 (UTC)

Done -- @Jeraphine Gryphon: Access granted on the basis of alternative account. West.andrew.g (talk) 16:36, 27 May 2015 (UTC)

Access

Hi. May I please have access to STiki. I have downloaded it. GLG GLG (talk) 17:46, 31 May 2015 (UTC)

Done -- @GLG GLG: On the basis of having 700+ article namespace edits and much anti-vandalism work. West.andrew.g (talk) 14:59, 1 June 2015 (UTC)

Possible "semi-emergency" release in the works

A Mediawiki API change is about to be brought into force: [1]. At the end of June, deprecated functionality involving continuation/paginated query results on the API will effectively be turned off. I need to read the documentation more closely and determine if, and/or to what extent, STiki is affected. This may result in a quick-release and forced-upgrade if impacts are significant. Also note per the [WikiTech-l] post that a large number of bots seem to be affected. I will applaud the WMF/Mediawiki for trying to broadly publicize this change, rather than leaving tool developers scrambling at the time of switchover. Thanks, West.andrew.g (talk) 21:05, 5 June 2015 (UTC)

I want access

I want to revert vandalism. I want my alt account Frænky.TG! (talk · contribs) to be granted permission to access STiki. -- FrankBoy _CHITCHAT 15:33, 8 June 2015 (UTC)

Done -- @FrB.TG: Approved on basis of alternate account. I'm hoping STiki doesn't choke on that special character. Thanks, West.andrew.g (talk) 11:56, 10 June 2015 (UTC)

STiki Access

I'm a new (1 day old, 35 edits) editor working on de-vandalizing the wiki. I'm assuming the answer will be a few more days, but just in case I'd like to check now. Thanks! (talk to) TheOtherGaelan('s contributions) 06:36, 10 June 2015 (UTC)

I guess you have the right answer... But after your first 35 edits you have been adding many more, so working in the right direction. With about a 1000 edits you will be welcome (as I would guess - but not up to me..).Super48paul (talk) 08:10, 10 June 2015 (UTC)

A "new account" approval isn't going to happen; we need a demonstrated track record of anti-vandalism/patrolling work. However, given this, I often approve accounts with far less than 1000 article namespace edits. I will also point out that I consider enrollment and subsequent completion of WP:CVUA to put accounts on the "fast track" to enrollment. Thanks, and we hope to see you again shortly. West.andrew.g (talk) 12:03, 10 June 2015 (UTC)

FYI, I have 35 article-space edits, but quite a few more non-articlespace edits. As a requirement for STiki is 1000 article-space edits, I presumed that article-space edits would be the statistic considered for talk-page approvals. (talk to) TheOtherGaelan('s contributions) 14:15, 10 June 2015 (UTC)

STiki access

Most of my edits since my account creation is cleaning up vandalism. I can easily distinguish vandalism between GF edits. If you can approve my usage of STiki I would appreciate West.andrew.g. Thanks very much. —Fazbear7891 (talk) 03:42, 12 June 2015 (UTC)

@Fazbear7891: You already have the rollback permission, so you should be able to use STiki already. --I am k6ka ^{Talk to me!} _{See what I have done} 16:51, 15 June 2015 (UTC)

Access

Hello! I am currently researching wikipedia's vandal fighting software. I have relatively few edits, but I would like to experiment with the STiki interface. May I please have access? Thank you! aetilley 04:36, 15 June 2015 (UTC) — Preceding unsigned comment added by Aetilley (talk • contribs)

With just 35 edits overall, it's highly unlikely. --Rsrikanth05 (talk) 20:04, 15 June 2015 (UTC)

Have responded to user on their talk page. Feel free to comment. --Rsrikanth05 (talk) 04:17, 16 June 2015 (UTC)

Error logging in

Hello @West.andrew.g: I have trouble logging into STiki. I'm receiving the following error; "Error in the user login interface, likely caused by network error Premature end of file." Would appreciate your help in this regard. Thanks. --JAaron95 (Talk) 16:23, 12 June 2015 (UTC)

STiki has been down for some hours now. Widr (talk) 16:32, 12 June 2015 (UTC)

Oh! Okay..

--JAaron95 (Talk) 16:38, 12 June 2015 (UTC)

For some reason I read it as End of Life and wondered whether the Stiki server was being shut down or not. --Rsrikanth05 (talk) 07:16, 13 June 2015 (UTC)

@Rsrikanth05: LOL!

--JAaron95 (Talk) 12:45, 13 June 2015 (UTC)

Safe to state that since it is the Weekend, Stiki would most likely be down till Monday. Unless the problem can be fixed remotely. --Rsrikanth05 (talk) 14:01, 13 June 2015 (UTC)

Also, I have created a template here: Wikipedia:STiki/Down that will display STiki is down. Feel free to use it on the project page. --Rsrikanth05 (talk) 14:09, 13 June 2015 (UTC)

-- The server is not down. This is an API change or WMF issue. Investigating now. West.andrew.g (talk) 18:35, 13 June 2015 (UTC)

THanks for the update. --Rsrikanth05 (talk) 19:54, 13 June 2015 (UTC)

I guess that explains why I'm only getting old edits with STiki after updating it. --I am k6ka ^{Talk to me!} _{See what I have done} 16:52, 15 June 2015 (UTC)

Well done, Andrew - seems to work again. Any explanation?Super48paul (talk) 18:28, 15 June 2015 (UTC)

STiki frontend is fixed (I believe). Still troubleshooting some related edit ingestion issues so people are probably seeing old edits. I'll do my best to have a fix tonight. This cascades across practically all my code, so there are also report generation and other issues to be cleaned up. I will report in full here later, but it essentially goes back to the fact that WMF Mediawiki installations can only use HTTPS now (not HTTP). Thanks, West.andrew.g (talk)

@West.andrew.g @Jaaron95 I am having the same issue on STiki. I can't connect to WP in both ST and HG. I am running on a mac. --Fazbear7891 (talk) 20:32, 16 June 2015 (UTC)