User talk:CheMoBot/Archive 1

Logging failure

Started seeing msgs on #wikichem:

<+CheMoBot> LOG - save error - log not saved to Wikipedia:WikiProject_Pharmacology/Log/2008-08-14.
<+CheMoBot> LOG - save error - log not saved to Wikipedia:WikiProject Chemicals/Log/2008-08-14.

just after the time it would have rolled over from logging 2008-08-13 to 2008-08-14. I created those two /Log/2008-08-14 pages manually, and logging resumed. From looking at the history of those pages and the -13 pages, seems like the bot lost its WP login credentials. As an anonIP user, it can't create new pages. So two issues: making sure it doesn't lose its login, and giving a more diagnostic message on #wikichem about why it failed to log. DMacks (talk) 03:54, 14 August 2008 (UTC)

Hmm, it should check its login before it saves, and if it does get logged-out should tell so. I'll amend the bot.

Please, next time use the settings to stop the bot from editing on. --Dirk Beetstra ^{T C} 15:38, 14 August 2008 (UTC)

Weird log annotation

This edit was noted on #wikichem as:

06:43 <+CheMoBot> user:Escarbot has edited chembox containing page Fructose-1-phosphate - http://en.wikipedia.org/w/index.php?diff=258072001&oldid=128908719 (+27) - Summary: ' robot  Adding: es:Fructosa-1-fosfato'
06:43 <+CheMoBot> Fructose-1-phosphate change: Removal of box (was Chembox) UNSET (8)

Why is it flagged as "Removal of box" DMacks (talk) 12:24, 15 December 2008 (UTC)

Weird log annotation

This edit was noted on #wikichem as:

13:16 <+CheMoBot> user:Benjah-bmm27 has edited drugbox containing page Heroin - http://en.wikipedia.org/w/index.php?diff=266352846&oldid=266346658 (+6) - Summary: ' new skeletal formula with up-to-date wedged hashed bonds'
13:16 <+CheMoBot> Heroin change: Removal of box (was drugbox) UNSET (8)

The only change was to the value of an already-existing field in the infobox. DMacks (talk) 18:23, 25 January 2009 (UTC)

Edit with oudated page

This edit [[1]] CheMoBot made appears to have used a version of the page at least one minute before (i.e. right before the vandalism was corrected). It thus restored the vandalism. I have restored the page to the intended version with the intended edit. --Multivariable (talk) 05:54, 8 September 2009 (UTC)

OK, that needs a tweak in the code, I will do that today. I know where the problem is, and I saw it earlier, but there thought 'what are the chances'. The problem is, the bot loads the current revid, and the verified revid, compares the data, adapts the chem/drugbox and saves. That takes a second or so, I would guess (but I don't know). But apparently it does happen too often that it 'edit conflicts'. I'll try to tweak ASAP. Thanks for telling! --Dirk Beetstra ^{T C} 06:17, 8 September 2009 (UTC)

I have put in a catch for this, but the active modules still need to restart to the new version (doing it manually would lose some edits, and I don't think that these rare circumstances need that). Again, thanks for keeping an eye on the bot. --Dirk Beetstra ^{T C} 07:41, 8 September 2009 (UTC)

Hmm, there was indeed ample time to edit conflict, it takes the bot 27 seconds to handle an edit to Resveratrol (that includes loading two full revids, extracting, updating, checking and rebuilding the page). No wonder that it sometimes edit-conflicts. --Dirk Beetstra ^{T C} 08:18, 8 September 2009 (UTC)

Use of internal database

Couldn't this bot use an internal database for its revid instead of filling my watch list and article history with meaningless edits?--Nutriveg (talk) 19:37, 9 September 2009 (UTC)

Using an internal database for the revids is very un-wikilike, and though we proposed that, it was not allowed. But I am not sure what edits you are referring to, are you referring to my burst of validations of about 160 pages with a drugbox, or to the mainspace edits by the bot? --Dirk Beetstra ^{T C} 21:13, 9 September 2009 (UTC)

I don't know how many they were but I often see that bot editing a page after someone else does just to change a revid, that way I can't directly check the recent changes through the "diff" link but need to access the article history. The bot should use its all means (like an own database) to keep the id of the wikipedia article version it last checked, not the article itself.--Nutriveg (talk) 21:56, 9 September 2009 (UTC)

I just checked the last 500 contributions of CheMoBot (earliest being 03:46, 8 September 2009). That's not enough to see what the bot is doing, but in that period there were six articles visited twice, and five of those were "good" (mainly restoring information that had been inadvertently removed while editors were reverting vandalism). Behavior on Glutathione was strange: it added {{cascite}} multiple times: [2] [3]. Johnuniq (talk) 04:02, 10 September 2009 (UTC)

Hmm, Nutriveg, I think you misunderstood what the bot is doing. The bot is checking, some time after an edit, if there are changes to the article it needs to be aware of, and it is setting flags in the articles.

Where missing, it adds <verifiedrevid> and <CASNo_Ref>, the first to signify the verified revid as set in the on-wiki index, the other to signify if the CAS-number is the same as in the verified revid. Both are needed to show readers if a page is changed. That can not be done by some off-wiki or elsewhere stored revid, as that would not result in a 'change' to the document. When <verifiedrevid> and/or <CASNo_Ref> are there, the bot generally does not 'follow the edit', but leaves the edit be (check the page history of e.g. Sodium hydroxide, it did one edit to add the verifiedrevid, after that there are three edits, but as nothing changed that was within the scope of the bot, it does not edit it again). However:

If an editor would change a field in the infobox of some data that is generally 'immutable' (e.g. the boiling point of water; that will always be 100 °C), and the 'change' flag is not set, then the bot will follow up after some time (typically, 10 minutes after the last edit) to set the flags. If after that the field is 'corrected', the bot will again follow up to remove such flags (if an editor does not remove them already).

I do realise, that this sometimes will result in the bot editing after a 'vandal' or bad edit. The bot will soon cease doing that, as more and more of the pages that are already verified have the verified revid. The delay of 10 minutes also results in most vandalism already being reverted before the bot edits (those that are not are indeed now one click further away, but can still be undone via undo on the previous edit). Soon, and that is our target, CheMoBot will start detecting a very difficult form of vandalism, where editors are changing one of the 'immutable' numbers into something else (do you know if an editor who is changing the boiling point of tri-n-butylphosphine from 35 to 75 °C is a) correcting it, b) vandalising it, or c) changing it in something else that is also incorrect?). Once those immutable numbers are verified in the indices, the bot will be able to tell you that, and you will, with one click, be able to see what is the verified number.

Regarding bugs, there were '3'are 4:

1. one where the 'regex' was set to 'greedy', so it added the same value multiple times. Resolved by setting the regex to not-greedy.
2. one that happened twice, once noticed by myself, second time noted by someone else (see post above). That was a form of 'edit-conflicting', where the bot started processing a page, and before it saved, someone else edited the page as well. After saving the 'old' version was incorporated. I have also resolved that bug by making the bot check if the revid it is working on is still the same as the one it is about to write over (which theoretically could still result in the edit conflict, but then it will be in the couple of millisecs between checking and saving, as opposed to the 6-7 seconds it takes to parse an edit).
3. one was a very strangely formatted chembox, something that did not really make sense, and hence, the bot did understandably not understand that, not really a bug, and would not know how to resolve that. It needs, in all cases, a human who repairs the wikipage (note: the bot just tagged the page, it did not 'break' the template or something)
4. the one that Johnuniq described, I have to look at that one later, that is curious, strange mistake.

I hope this explains. --Dirk Beetstra ^{T C} 06:20, 10 September 2009 (UTC)

I solved the fourth bug. --Dirk Beetstra ^{T C} 14:42, 10 September 2009 (UTC)

Bug

https://secure.wikimedia.org/wikipedia/en/w/index.php?title=Potassium_dichromate&diff=318294924&oldid=318294914 Verbal chat 18:38, 6 October 2009 (UTC)

HUH! This was an old bug and I solved it .. This must have been split second work. Let me explain:

1. the bot takes the last edit time of the page, approximately 10 minutes after the last edit.
2. the bot loads the page.
3. the bot parses the box, this takes several seconds.
4. the bot again takes the last edit time of the page.
5. next line of code, the bot checks the two times, if they are the same, no-one edited inbetween, and proceeds, otherwise it re-queues and starts with step 1
6. next line of code: if the bot proceeds, it saves the new pages IMMEDIATELY after the last step.

This edit suggests that in those milliseconds that it takes to push the data through the network connection onto the server, you managed to revert the vandalism. I'm not sure if this is catchable or solveable, though it is a bug, indeed. I'll give it another think, but I am afraid that this is a rare occasion. --Dirk Beetstra ^{T C} 19:22, 6 October 2009 (UTC)

This seems to be the same thing. Just wanted to let you know there are at least two instances of this bug. --ἀνυπόδητος (talk) 19:56, 26 October 2009 (UTC)

This bugs me .. :-) .. I will have to try and solve this problem in some way or another. Not sure how yet. --Dirk Beetstra ^{T C} 20:29, 26 October 2009 (UTC)

Screwing up of the Drugbox

http://en.wikipedia.org/w/index.php?title=Methadone&diff=390221074&oldid=390219825

The CheMoBot revision destroyed the drugbox syntax by putting in "$1|FDA}}" where it makes no sense. 60.242.66.148 (talk) 09:11, 12 October 2010 (UTC)

Indeed. That makes no sense. I have reverted, now looking at the bot to see what it does wrong. Thanks!! --Dirk Beetstra ^{T C} 09:31, 12 October 2010 (UTC)

I think I solved it. Thanks for reporting this. I hope this bug did not screw up too many boxes (it has been active since yesterday, I'll check). --Dirk Beetstra ^{T C} 10:00, 12 October 2010 (UTC)

http://en.wikipedia.org/w/index.php?title=Creatine&diff=401482370&oldid=401480730

http://en.wikipedia.org/w/index.php?title=Creatine&diff=401482370&oldid=401480730

Eh .. it should do that .. but then with 'correct' or 'changed' in the template ... thanks for noting this, looking at it. --Dirk Beetstra ^{T C} 10:12, 10 December 2010 (UTC)

minus/ndash Suggestion

In changing Psoralen, I changed | MeltingPt = 158-161 °C so the range used an ndash. This triggered a watched field changed update. You may wish to ignore changes of minus to and from ndash if that is easy (something like) /\b(-|–|–)\b/ RDBrown (talk) 06:50, 17 December 2010 (UTC)

That is a useful suggestion, did not think about those things. I will write a function for that in the bot (for such string comparisons), and use User:CheMoBot/Settings for feeding the possibilities. Have to think though how to do it exactly.

• standardize_string=dash
• standardize_dash_regex=\b(-|–|&ndash;)\b
• standardize_dash_standard=-

Where the code replaces anything matching the regex out of the first param with what is 'the standard' on both strings to compare, and then compares the end results and returns 1 when equal .. Sounds like a possibility.

Thanks, I will do this after X-mas/New Year. Talking about that, have a Merry Christmas and a Happy 2011! --Dirk Beetstra ^{T C} 08:00, 17 December 2010 (UTC)

Actually, implemented and testing now, see bottom of User:CheMoBot/Settings. Thanks again. --Dirk Beetstra ^{T C} 08:12, 17 December 2010 (UTC)

OrphanReferenceFixer: Help on reversion

Hi there! I check pages listed in Category:Pages with incorrect ref formatting to try to fix reference errors. Recently, you reverted my fix to Simvastatin.

If you did this because the references should be removed from the article, you have misunderstood the situation. Most likely, the article originally contained both <ref name="foo">...</ref> and one or more <ref name="foo"/> referring to it. Someone then removed the <ref name="foo">...</ref> but left the <ref name="foo"/>, which results in a in the article. I replaced one of the remaining <ref name="foo"/> with a copy of the <ref name="foo">...</ref>; I did not re-insert the reference to where it was deleted, I just replaced one of the remaining instances. What you need to do to fix it is to make sure you remove all instances of the named reference so as to not leave any .

If you reverted because I made an actual mistake, please be sure to also correct any reference errors in the page so I won't come back and make the same mistake again. Also, please post an error report at User talk:AnomieBOT so my operator can fix me! If the error is so urgent that I need to be stopped, also post a message at User:AnomieBOT/shutoff/OrphanReferenceFixer. Thanks! AnomieBOT⚡ 18:05, 10 June 2011 (UTC) If you do not wish to receive this message in the future, add {{bots|optout=AnomieBOT-OrphanReferenceFixer}} to your talk page.

An odd removal

Here. 22:28, 19 July 2011 (UTC)

Edit conflict .. solved below. SORRY. --Dirk Beetstra ^{T C} 07:58, 30 August 2011 (UTC)

Please don't edit the drugbox documentation

I think this isn't a good idea. Even reverting does not help. Could you keep CheMoBot from editing Template:drugbox/doc? Thanks, ἀνυπόδητος (talk) 09:36, 21 July 2011 (UTC)

This page should not be indexed in Wikipedia:WikiProject Pharmacology/Index. Then the bot will leave it alone. --Dirk Beetstra ^{T C} 07:59, 30 August 2011 (UTC)

Edit summary tip

If you want to save a few characters in the edit summary, try using a shortcut (WP:CHEMVALID, or similar) to the validation page instead of the full path (Wikipedia:WikiProject Chemicals/Chembox validation Headbomb {talk / contribs / physics / books} 20:04, 21 August 2011 (UTC)

Good idea! --Dirk Beetstra ^{T C} 07:47, 30 August 2011 (UTC)

Should be done now .. it is actually a setting. --Dirk Beetstra ^{T C} 07:59, 30 August 2011 (UTC)

Reverting edits outside infobox

I thought this bot was only supposed to verify facts in an {{Infobox}} or {{Drugbox}}. Why did it revert my formatting edit that was outside of all such boxes? And why revert (which is brash) instead of marking the edited content as unverified (thus assuming good faith)? —voidxor (talk | contrib) 06:13, 30 August 2011 (UTC)

Eh .. I think this needs a big SORRY. The bot is not reverting, it is 'edit conflicting'. The bot loads a page, extracts the infobox, does parsing (which takes a bit of time), and saves. I will work on this again, apparently there is still too much time between the doublecheck and the actual save (it is a recurring issue, see above as well). --Dirk Beetstra ^{T C} 07:47, 30 August 2011 (UTC)

Ehm

Ehm ...

I should be hiding under a big rock .. this 'bug' was reported before, and I 'repaired' it. Problem is, it is a difficult bug to test. Problem was .. there was a minor thingy wrong in my fix, which made it not work. Sorry again, I will restart the handlers, it is fixed now. SORRY again. --Dirk Beetstra ^{T C} 07:58, 30 August 2011 (UTC)

I reverted the revert and all is well now. Thanks for the explanation and for fixing it. —voidxor (talk | contrib) 06:43, 31 August 2011 (UTC)

Chembl ids

Please take a look at the revision history of arginine. The version of 27 July 2011 contains the Chembl id of 179653, which is incorrect, yet had a green check mark beside it. On August 2 a user changed the id to 1485, which is correct. Then the CheMoBot came in and added a red cross next to the correct id. There are numerous such cases to be found in the contribution history of user:Louisajb. Couldn't the bot determine whether an id is (likely) correct? Cheers, AxelBoldt (talk) 20:52, 18 September 2011 (UTC)

Eh, no. The verified revid has to be set in the index, you could correct that, depending on the box in Wikipedia:WikiProject Chemicals/Index or Wikipedia:WikiProject Pharmacology/Index, the bot will then, soonish, take care of it. Thanks! --Dirk Beetstra ^{T C} 14:04, 14 October 2011 (UTC)

Blanking

Blanking here.  -- WikHead (talk) 01:21, 11 November 2011 (UTC)